Seminar Schedule

Intel Labs Seattle Seminars occur on Wednesdays from 4 to 5pm (unless otherwise noted) at the Intel Labs Seattle office (location information), and are open to the research community. We are pleased to have you join us for light refreshments and tea prior to the event to meet the guest speaker and audience. Most seminars are also recorded and available as video-on-demand after the talk--see the links with each past seminar entry below.

Upcoming Seminars

Name: Ryan Nikolaidis, Georgia Tech
Date: Friday, September 3
Time: 10:00 - 11:00 a.m..
Host: Emily Cooper
Title: Music You Can Click Your Mouse To
Abstract: My research spreads across fields as varied as music intelligence, improvisation, sonification, and brain research. In common, all of these studies rely on generative systems. However, each presents unique obstacles that demand unique solutions. I offer a glance at each of these areas, how their challenges lead to exciting new possibilities, and how they influence each other and future research.
Bio: Ryan Nikolaidis is a PhD student and researcher in Music Technology at the Georgia Institute of Technology. His current research focuses on generative music systems. Applying this research has led him to work in fields including robotic musicianship and dynamic display sonification. His interests include human-inspired theoretical cognitive models and machine learning for creative purposes. Currently, he has started a series of brain-imaging studies to investigate how we perceive and create musical gestures. From Stetson University he received a BA in Music, with majors in guitar performance, composition, and music theory. His compositional studies included instruction from world-renowned composer Sydney Hodkinson. From 2000-06 he also studied classical guitar with distinguished performer and professor Stephen Robinson. He followed that education with post-baccalaureate work in Electrical Engineering at the University of Florida, ultimately leading to a blending of the two fields in Music Technology.

Past Seminars

Name: Wolfram Burgard, University of Freiburg
Date: Monday, Aug 23
Time: 3:00 - 4:00 p.m.
Host: Dieter Fox
Title: Graph-based SLAM: How to extend it for incorporating prior knowledge and how to use it for the evaluation of SLAM algorithms
Abstract: In this presentation I will first talk about how to extend graph-based approaches to the simultaneous localization and mapping (SLAM) problem towards the incorporation of background information. Most existing solutions to the SLAM problem rely on loop-closures to obtain global consistency and do not exploit prior information even if it is available. I will present one approach that improves SLAM by utilizing publicly accessible aerial photographs as prior information. Our approach inserts correspondences found between three-dimensional laser range scans and the aerial images as constraints into the underlying constraint graph. I will present results obtained on large real-world datasets acquired in a mixed in- and outdoor environment that clearly indicate the advantages of our method. Additionally, I will describe how a similar approach can be applied to improve global consistency in situations in which multi-story buildings are mapped with multiple robots. Finally, I will talk about the more fundamental question about the evaluation of SLAM algorithms and propose a technique that is based on the graph-based formulation of the SLAM problem. I will discuss this measure also in relation with alternative ones and will present extensive results.
Bio: Wolfram Burgard studied Computer Science at the University of Dortmund and received his Diploma Degree in 1987. In 1990, he became a member of the University of Bonn where he got his Ph.D.~in 1991. From 1991 to 1999 Wolfram Burgard was a PostDoc at the University of Bonn. Since 1999, Wolfram Burgard is a professor at the University of Freiburg where he heads the Laboratory for Autonomous Intelligent Systems (AIS). His major research interests include mobile robotics, state estimation and control, and artificial intelligence. In his career, Wolfram Burgard has published two books and over 200 papers and articles in outstanding journals and conference proceedings. For his scientific contributions he has received eight best paper awards
from major conferences including the IEEE International Conference on Robotics and Automation (ICRA), IEEE International Conference on Intelligent Robots and Systems (IROS), and the National Conference on Artificial Intelligence (AAAI). Wolfram Burgard is an active member of the IEEE Robotics and Automation Society as well as a life-time member of the American Association of Artificial Intelligence (AAAI). He is Fellow of the European Coordination Committee for Artificial Intelligence (ECCAI) and the Association for the Advancement of Artificial Intelligence (AAAI). In 2009, Wolfram Burgard received the Gottfried Wilhelm Leibniz Prize of the German Research Foundation (DFG), the most prestigious German research award.
Talk

Name: Philipp Slusallek, German Research Center for Artificial Intelligence (DFKI)
Date: Friday, Aug 6
Time: 10:00 - 11:00 a.m..
Host: Aaron Lefohn
Title: Intelligent Simulated Reality in the 3D Internet
Abstract: The goal of the Intelligent Simulated Reality (IsReal) Project is to bring together research from graphics and AI to create large-scale, detailed, and realistic models of the real world. Semantic annotations and reasoning approaches of different forms allow meaningful interactions of virtual agents with their environment but also support the interaction of a human user with the virtual environment. To make the system available to a broad range of users we have implemented fully interactive 3D graphics for Web browsers using a declarative scene description (XML3D) that is easily programmable by any Web developer. In my talk I will present some of our current research activities and discuss key research challenges, including user interaction in a future 3D Internet.
Bio: Philipp Slusallek is Scientific Director at the German Research Center for Artificial Intelligence (DFKI) in Saarbrücken, where he leads the research area “Agents and Simulated Reality”. In 2009 he was also appointed Director of Research at the Intel Visual Computing Institute of Saarland University, where he has been full professor for computer graphics since 1999. Before joining Saarland University he was visiting assistant professor at the Stanford University Graphics Lab. He received a Diplom (M.Sc.) in Physics from Tübingen University and a Dr.-Ing.(PhD) from the University of Erlangen in Germany. His research interests include the integration of Graphics and Artificial Intelligence, Future 3D Internet and Digital Media, High-Performance Visual Computing and Simulation, High-Quality Interactive Rendering, Parallel Programming Models and Compiler, and Adaptive 3D & Immersive User Interfaces.
Talk

Name: Prabal Dutta, University of Michigan
Date: Wednesday, Jul 28
Time: 4:00 - 5:00 p.m.
Host: Ben Greenstein
Title: Pervasive Personal Sensing
Abstract: This talk will present our recent work on enabling pervasive personal sensing, focusing on HiJack, a system for stealing power and bandwidth from the mobile phone's audio jack. HiJack enables a new tier small and cheap phone-centric sensor peripherals that support plug-and-play operation. More broadly, we envision the mobile phone will become a portal for perpetually-powered and physically-embedded sensors, so this talk will discuss augmenting indoor environments with energy-harvesting sensors, and bridging the cyber and physical worlds with mobile phones and streaming web applications.
Bio: Prabal Dutta is an Assistant Professor at the University of Michigan, Ann Arbor. His interests span systems, networking, and architecture, with a focus on wireless, embedded, networked, mobile, and sensing systems. His work has been commercialized by Aginova, Arch Rock, Crossbow, Moteiv, Moteware, SonnOnet, and Vectare, and is in use by hundreds of researchers and practitioners worldwide (including Intel Labs Berkeley). He earned a Ph.D. in Computer Science from the University of California, Berkeley.

Name: Romit Roy Choudhury, Duke University
Date: Wednesday, Jul 21
Time: 4:00 - 5:00 p.m.
Host: Anmol Sheth
Title: Designing a Virtual Information Telescope Using Mobile Phones and Social Participation
Abstract: Modern phones are being equipped with numerous sensors such as cameras, microphones, GPS, accelerometers, and health monitors. This project aims to design a "Virtual Information Telescope", where the "lenses" of the telescope are metaphors for the sensors in people's mobile phones. Using such a telescope, an Internet user will be able to zoom into any part of the human-populated world, and observe events of interest. Users will be able to direct queries to phones located in a given region, and receive real-time responses through automatic sensing or explicit human participation. Example domains that may benefit from this platform include education, healthcare, tourism, disaster management, environment conservation, and social collaborations. Perhaps more fundamentally, a virtual information telescope may change the way we browse, query, learn, and process information. This talk will expand on this vision, and instantiate it through a live system called "Micro-Blog". We will discuss a suite of important research challenges underlying the translation of Micro-Blog into a deployable/usable system. Of particular interest are topics on energy-efficient localization, symbolic localization, information distillation, location-privacy, and user interfaces. We will close not only with challenges we are struggling with, but will also look forward to what may lie ahead along the path of mobile, social, participatory computing.
Bio: Romit Roy Choudhury is an Assistant Professor of ECE and CS at Duke University (he is spending the summer as a visiting researcher at Microsoft Research, Redmond). He joined Duke in Fall 2006, after completing his PhD from UIUC. His research interests are in wireless protocol design mainly at the PHY/MAC layer, and in mobile social computing at the application layer. He received the NSF CAREER Award in January 2008. Visit Romit's Systems Networking Research Group (SyNRG), at http://synrg.ee.duke.edu
Talk

Name: Neel Joshi, Microsoft Research
Date: Tuesday, Jul 6
Time: 10:00-11:00 a.m.
Host: Xiaofeng Ren
Title: CSI in Real Life? New Technologies for Image Enhancement
Abstract: Have you ever snapped a photo of memorable event, only to find that you didn’t quite get the result you really wanted? Was the photo too blurry, too noisy, or did it just not convey the moment? In this talk, I will discuss recent work that improves the quality of your photos using image enhancement techniques and also discuss new methods for image content creation. Our recent work in single image enhancement can automatically reduce image blur and noise and can be used to tag and filtering photos with their blur and noise level. A second method uses a novel hardware attachment and a blur-estimation algorithm to deblur images automatically. With traditional photography, we are all used to taking a single photograph of a scene. However, a photographer’s intent is often to capture more than what can be seen in a single photograph. I will additionally present work that combines hundreds or even thousands of images to create images that are much better than a single photograph.
Bio: Neel Joshi has been taking a lot of bad photos (and a few good ones) ever since he got his hands on a camera at age five. He is currently a Postdoctoral Researcher in the Interactive Visual Media Group at Microsoft Research, working in computer vision and graphics. Neel earned an Sc.B. from Brown University, an M.S. from Stanford University, and a Ph.D. in computer science from U.C. San Diego. He held internships at Mitsubishi Electric Research Labs, Adobe Systems, and Microsoft Research. Neel was recently a visiting professor at the University of Washington.
Talk

Name: Piotr Dollar, Caltech
Date: Thursday, July 1
Time: 11:00-12:00 p.m.
Host: Xiaofeng Ren
Title: Pedestrian Detection: The State of the Art
Abstract: Pedestrian detection is a key problem in computer vision, and truly accurate pedestrian detection would have immediate and far reaching impact on areas such as robotics, surveillance, assistive technology for the visually impaired, image indexing, advanced human machine interfaces and automotive safety, among others. In the first part of the talk I will discuss the state of the art in monocular pedestrian detection, including our large-scale benchmarking effort, highlighting current successes and challenges for the research community. In the second part of the talk I will describe our own pedestrian detection approach which achieves top performance in numerous settings across multiple datasets. In particular, I will focus on a recent insight that has allowed us to perform multiscale detection at near real time on standard hardware. Our approach yields a speedup of 10-100 times over competing methods with only a minor loss in detection accuracy, and the underlying theory should be readily applicable to other domains. Finally, I will conclude by discussing open challenges in pedestrian detection and ideas on how to address them.
Bio: As of September 2007, I am a postdoc at the Computation Vision lab at Caltech under Pietro Perona. My graduate work was under the guidance of Serge Belongie and supported by the NSF IGERT fellowship. I successfully defended my PhD in Computer Science at UCSD in August 2007; the topic of my dissertation was "Learning from Local Image Regions." My interests lie in machine learning and pattern recognition, and their application to computer vision. http://vision.ucsd.edu/~pdollar/
Talk

Name: Eyal de Lara, University of Toronto
Date: Friday, June 4, 2010
Time: 10:00 - 11:00 a.m.
Host: Anmol Sheth
Title: SnowFlock: Rapid Virtual Machine Cloning for Cloud Computing
Abstract: A major advantage of cloud computing is the ability to use a variable number of physical machines and virtual machine (VM) instances depending on the needs of the problem. Unfortunately, instantiating new VMs on existing clouds, such as Amazon's EC2, is a slow operation that typically takes "minutes." This model's lack of agility fails to provide users the full potential of the cloud, and forces application providers to overprovision, thus wasting valuable resources. In this talk, I will describe VM fork, a new abstraction that can replicate a VM into hundreds of cloud hosts in less than a second. While VM fork is conceptually a simple idea, the large size of VM memory images (which can reach several gigabytes) makes an efficient implementation challenging. SnowFlock, our implementation of the VM fork abstraction, addressed this problem by leveraging two keyobservations: a child VM is likely to access just a small part of the parent's memory image; and the significant locality that exists between the memory accesses of the children VMs of a common parent VM. Based on this intuition, SnowFlock includes three key mechanisms: Lazy State Replication, which allows for extremely fast instantiation of VM clones by initially copying the minimal necessary data to resume a VM on a remote host, and transmitting only the fraction of the parent's state that the clone actually accesses on demand; Avoidance Heuristics, which eliminate superfluous memory transfers for the common case of clones allocating new private state; and Multicast Distribution, a technique that provides scalability and pre-fetching by multicasting replies to memory page requests.
Bio: Eyal de Lara is an Associate Professor in the Department of Computer Science at the University of Toronto. Eyal received his Ph.D. and M.Sc.from Rice University in 2002 and 1999, and a B.Sc. from the Instituto Tecnologico de Monterrey in 1995. His research interests include distributed systems and mobile computing. His research has been recognized with an IBM Faculty Award and a NSERC Discovery Accelerator Award.
Talk

Name: Derek Nowrouzezahrai, University of Toronto
Date: Monday, May 17
Time: 10:00 - 11:00 a.m.
Host: Dieter Fox
Title: Rendering: an Artist's Nightmare (and a Researcher's Dream)
Abstract: This talk will begin with a brief discussion of rendering at a high-level, including where it fits in the bigger picture of the CG pipeline. Those unfamiliar to the area will be given a quick tour of the field, as well as the diverse mathematical and computational approaches used to solve different rendering problems. Moreover, this talk will discuss how seemingly unrelated areas of computer science influence the process of generating realistic images. Lastly, this talk will motivate and elaborate data-driven, basis space solutions to real-time rendering problems, and how this sub-area of rendering is poised to change the way digital artists generate content for the next generation of feature films and interactive entertainment productions. In turn, the caliber of this content, as well as the degree to which it can be manipulated interactively, will grow significantly in the near future.
Bio: Derek Nowrouzezahrai is a PhD candidate at the University of Toronto, completing his studies under the supervision of Dr. Eugene Fiume. His primary research interests revolve around developing accurate approaches for real-time rendering in dynamic environments. In addition to his rendering interests, Derek has also contributed to the areas of fluid animation, geometry processing, and real-time non photorealistic rendering. Derek completed his MSc at the University of Toronto in 2006, and his BASc at the University of Waterloo in 2005. Derek was an intern at Microsoft Research under the supervision of Dr. John Snyder, as well as at Disney Research Zurich under the supervision of Dr. Wojciech Jarosz, and has held prior affiliations with Electronic Arts Canada, Research in Motion, Amazon.com, and Microsoft Corporation.
Talk

Name: David Wingate, MIT
Date: Tuesday, May 11
Time: 10:00 - 11:00 a.m.
Host: Dieter Fox
Title: Hierarchical Bayesian Methods for Reinforcement Learning
Abstract: Designing autonomous agents capable of coping with the complexity of the real world is a tremendous engineering challenge. Such agents must often deal with rich observations (such as images), unknown dynamics, and rich structure---perhaps consisting of objects, their properties/types and their dynamical interactions. An ability to learn from experience and generalize radically to new situations is essential; at the same time, the agent may bring substantial prior knowledge to bear on the environment it finds itself in. In this talk, I will present recent work on the combination of reinforcement learning and nonparametric Bayesian modeling. Hierarchical Bayes provides a principled framework for incorporating prior knowledge and dealing explicitly with uncertainty, while reinforcement learning provides a framework for making sequential decisions under uncertainty. I will discuss how nonparametric Bayesian models can help answer two questions: 1) how can an agent learn a representation of state space in a structured domain? and 2) how can an agent learn how to search for good control laws in hard-to-search spaces? I will illustrate the concepts on applications including modeling neural spike train data, causal sound source separation and optimal control in high-dimensional, simulated robotic environments.
Bio: David Wingate received a B.S. and M.S. in Computer Science from Brigham Young University in 2002 and 2004, and a Ph.D. in Computer Science from University of Michigan in 2008. He is currently a postdoctoral research associate in the Computational Cognitive Science group at MIT.
David's research interests lie at the intersection of perception, control and cognition. His research spans diverse topics in reinforcement learning, Bayesian unsupervised learning, information theory, manifold learning, kernel methods, massively parallel processing, visual perception, and optimal control.
Talk

Name: David Hsu, NUS
Date: Monday, May 10
Time: 2:00 - 3:00 p.m.
Host: Dieter Fox
Title: A POMDP Approach to Robot Motion Planning under Uncertainty
Abstract: Motion planning in uncertain and dynamic environments is critical for reliable operation of autonomous robots. Partially observable Markov decision processes (POMDPs) provide a powerful framework for such planning tasks and have been successfully applied to several moderately complex robotic tasks, including navigation, manipulation, and target tracking. The challenge now is to scale up POMDP planning algorithms and handle more complex, realistic tasks. I will outline some ideas aimed at overcoming two major obstacles to the efficiency of POMDP planning: ``curse of dimensionality'' and ``curse of history''. Our main objective is to show that using these ideas---along with others---POMDP algorithms can be used successfully for motion planning under uncertainty for robotic tasks with a large number of states or a long time horizon. I will also highlight some challenges ahead.
Bio: David Hsu is currently an associate professor of computer science at the National University of Singapore and a member of NUS Graduate School for Integrative Sciences & Engineering (NGS). His research spans robotics, computational biology, and geometric computation. His current interest includes robot motion planning under uncertainty. He received B.Sc. in computer science & mathematics from the University of British Columbia, Canada and Ph.D. in computer science from Stanford University, USA. After leaving Stanford, he worked at Compaq Computer Corp.'s Cambridge Research Laboratory and the University of North Carolina at Chapel Hill. At the National University of Singapore, he held the Sung Kah Kay Assistant Professorship and was a Fellow of the Singapore-MIT Alliance
Talk

Name: Jim Rehg
Date: Monday, May 10
Time: 12:00 - 1:00 p.m.
Host: Dieter Fox
Title: Recognition of Actions in Egocentric Video
Abstract: We address the problem of action recognition in egocentric video--- video collected with a wearable camera that provides a “first person” view of the scene in front of the user. Such video provides a unique opportunity for large-scale, continuous capture of daily activities in natural settings (e.g. the activities of daily living.) Activities that are captured under these conditions are typically built-up from actions that involve the manipulation of objects, such as opening jars, drinking from a glass, etc. We present an approach to action recognition which leverages domain knowledge and context as derived from the egocentric paradigm. In particular we demonstrate that context-driven bottom up segmentation can yield semantically-meaningful features for action recognition. We show that the use of such semantic features can yield favorable results in comparison to a standard bag of words model. We believe these are the first action recognition results within the egocentric paradigm, based on visual features alone. This project is part of a larger effort in using computer vision technologies to support the detection, treatment, and understanding of developmental disorders such as autism.
This is joint work with Alireza Fathi and Xiaofeng Ren.
Bio: Jim Rehg is a Professor in the School of Interactive Computing at the Georgia Institute of Technology. He is co-Director of the Computational Perception Lab and Associate Director of Research in the Center for Robotics and Intelligent Machines. He received his Ph.D. from CMU in 1995 and worked at the Cambridge Research Lab of DEC (and then Compaq) from 1995-2001, where he managed the computer vision research group. His research interests include computer vision, robotics, and machine learning. He recently served as the general co-chair for CVPR 2009 in Miami.

Name: Mohit Gupta, CMU
Date: Wednesday, April 28
Time: 2:00 - 3:00 p.m.
Host: Xiaofeng Ren
Title: Probing Scenes with Programmable Illumination
Abstract: Projectors are dual devices to cameras. They project 2D images on 3D scenes in much the same way as cameras map the 3D world on 2D images. By using this duality, projectors can be treated as extremely versatile and programmable illumination sources. We can achieve high-resolution and high-speed modulation of illumination, change focus, zoom and wavelength of projected light. Projectors thus provide a large degree of control over interaction of light with scenes, and consequently, the resulting images. I will talk about scenarios where such programmable illumination can be used as a probe to recover a variety of scene properties, often in quite challenging conditions. In particular, I will present our work on recovering high-resolution depth-maps in the presence of inter-reflections and sub-surface scattering. Second, I will talk about designing active vision systems which can see clearer and farther in poor visibility conditions such as underwater scenarios, bad weather and smoke. Finally, I will show how programmable illumination can be used for designing high-speed motion-aware cameras. Such cameras would allow changing the spatio-temporal resolution of videos post-capture depending on the scene content.
Bio: Mohit Gupta is a fifth year PhD student in the Robotics Institute, CMU advised by Srinivasa G. Narasimhan. His research interests are in global light transport, computational imaging and illumination and
projector-camera systems. Prior to CMU, he received his B.Tech. in Computer Science from the Indian Institute of Technology, New Delhi in 2003 and an M.S. in Computer Science from the Stony Brook University in 2005.
Talk

Name: Daniel Vlasic, MIT
Date: Monday, April 26
Time: 10:00 - 11:00 a.m.
Host: Xiaofeng Ren
Title: Reconstruction and Analysis of Dynamic Shapes
Abstract: Motion capture has revolutionized entertainment and influenced fields as diverse as the arts, sports, and medicine. This is despite the limitation that it tracks only a small set of surface points. On the other hand, 3D scanning techniques digitize complete surfaces of static objects, but are not applicable to moving shapes. I present methods that overcome both limitations, and can obtain the moving geometry of dynamic shapes (such as people and clothes in motion) and analyze it in order to advance computer animation. Further understanding of dynamic shapes will enable various industries to enhance virtual characters, advance robot locomotion, improve sports performance, and aid in medical rehabilitation, thus directly affecting our daily lives. My methods efficiently recover much of the expressiveness of dynamic shapes from the silhouettes alone. Furthermore, the reconstruction quality is greatly improved by including surface orientations (normals). In order to make reconstruction more practical, I strive to capture dynamic shapes in their natural environment, which I do by using hybrid inertial and acoustic sensors. After capture, the reconstructed dynamic shapes are analyzed in order to enhance their utility. My algorithms then allow animators to generate novel motions, such as transferring facial performances from one actor onto another using multilinear models, or transferring full-body movements between different characters using only a few examples. The presented research provides some of the first and most accurate reconstructions of complex moving surfaces, and is among the few available approaches that establish a relationship between different dynamic shapes.
Bio: Daniel Vlasic grew up in Croatia. I came to MIT in 1997 as an undergrad andstayed for Masters and PhD. My thesis, supervised by Prof. Jovan Popovic, deals with capturing and animating shapes that deform.
Talk

Name: Nicholas Roy, MIT
Date: Friday, April 23
Time: 10:00 - 11:00 a.m.
Host: Dieter Fox
Title: Bridging the Gap Between Machine and Human Representations
Abstract: In the last few years, how robots understand the world around them has advanced considerably. Examples include the autonomous vehicles in the DARPA Grand Challenges and Urban Challenge, the considerable work in robot mapping, and the growing interest in home and service robots. However, these example technologies and systems are still mostly restricted to research prototypes. One obstacle to getting more widely useful robots is that the way robots reason about their world is still pretty different to how people reason. Robots think in terms of point features, dense occupancy grids and action cost maps. People think in terms of landmarks, segmented objects and tasks (among other representations). There are good reasons why these are different, and robots are unlikely to ever reason about the world in the same way that people do. But, there has been recent work in bridging the gap between low-level geometry and control, and higher-level semantic representations. I will talk about this work, what the open challenges are and how addressing these challenges creates more capable robots.
Bio: Nicholas Roy is an Associate Professor in the Department of Aeronautics & Astronautics at the Massachusetts Institute of Technology and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT. He received his Ph. D. in Robotics from Carnegie Mellon University in 2003. His research interests include autonomous systems, mobile robotics, human-computer interaction, decision-making under uncertainty and machine learning.
Talk

Name: Andy Wilson, MSR
Date: Wednesday, April 7
Time: 4:00 - 5:00 pm
Host: Beverly Harrison
Title: Surface Computing and Beyond
Abstract: What started as a modest incubation effort has grown into the Surface Computing group at Microsoft. Surface, its first product, is but one example of an exciting category of form factors and user experiences. In this talk I would like to present a number of research projects that share the Surface Computing vision. Firstly, we have experimented with different form factors. For example, PlayAnywhere is a compact tabletop projection-vision system which brings Surface interactions to everyday surfaces, while TouchLight combines a transparent projection screen material with computer vision techniques, and FourBySix allows multiple designers to gather around a large-format Surface. We’ve even brought Surface technology to spherical and dome projection displays. We are also examining Surface interactions that go beyond what is expected of the traditional point cursor model. For example, Surface input may be embedded in a gaming physics simulation to obtain realistic manipulations based on simulated friction and collisions. Finally, I will describe how we have applied recently developed range-sensing cameras to enable interactions above the surface. All of these systems have the potential of changing the way we relate to computing, but they also pose serious challenges because they are so different from today's desktop computing systems.
Bio: Andy Wilson is a Senior Researcher at Microsoft Research. There he has been applying sensing technologies to enable new styles of human-computer interaction. His interests include gesture-based interfaces, computer vision, inertial sensing, display technologies and machine learning. In 2002 he helped found the Surface Computing group at Microsoft. Before joining Microsoft, Andy obtained his BA at Cornell University in 1993, and PhD at the MIT Media Laboratory in 2000. Publications and videos of his work are located at http://research.microsoft.com/~awilson.

Name: Timothee Cour, INRIA in Paris, in the Computer Vision and Machine Learning Willow lab
Date: Monday, March 3
Time: 10:00 - 11:00am
Host: Xiaofeng Ren
Title: Weakly supervised learning for multimodal video understanding
Abstract: The exponential growth of image datasets and online videos presents both a challenge and an opportunity for vision based semantic search and indexing. The amount of labeled data and processing power grows at a much slower pace, posing a difficulty for traditional, heavily supervised learning methods. I will present in this talk scalable, weakly supervised algorithms we developed for video understanding and object recognition, with a special focus on identifying people in TV shows and photo collections. Key components of the algorithms we present are (1) alignment between multiple modalities: images, audio and text, and (2) unified convex formulation with strong theoretical guarantees for learning under weak supervision. We consider a common scenario in many image and video collections, where only partial access to labels is available. For example, photo collections often contain several faces per image and a caption that only specifies who is in the picture, but not which name matches which face. Likewise, screenplays can tell us who is in a given movie scene, but not when and where they are on the screen. Our goal in each case is to learn a person classifier that disambiguates the labels of the training faces, and also generalizes to unseen data. We cast the task as a partially-supervised multiclass classification setting where each detected face is labeled ambiguously with more than one label. We propose a convex formulation based on minimization of a surrogate loss, and show theoretically and empirically that effective learning is possible even when all examples are ambiguously labeled. We also investigate the challenging scenario of naming people in videos without a screenplay. Our only source of 'supervision' are person references mentioned in dialog, such as "Hey, Jack!''. We resolve identities by learning a classifier incorporating multi-instance constraints from dialog, gender and local grouping constraints, in a unified convex formulation. Grouping constraints are provided by a novel temporal grouping model that learns a partition classifier from a set of training videos using structured learning. We have applied our algorithms on hundreds of hours of video, and will report quantitative and qualitative results. I will also briefly mention ongoing work on learning from weakly annotated image datasets.
Bio: Timothee Cour is a postdoctoral researcher at INRIA in Paris, in the Computer Vision and Machine Learning Willow lab. He graduated from Ecole Polytechnique in 2003 in Applied Mathematics and Computer Science, and obtained a MS and PhD in Computer Science from University of Pennsylvania in 2005 and 2009. His main research area lies at the intersection of Computer Vision and Machine Learning. He has recently worked on multimodal video understanding, object detection, activity recognition, image segmentation and graph matching, using scalable weakly supervised learning and approximate inference.
Talk

Name: Li Guan, University of North Carolina-Chapel Hill
Date: Monday, January 25
Time: 10:00 - 11:30am
Host: Xiaofeng Ren
Title: Multi-sensor Dynamic Scene Modeling
Abstract: Modeling dynamic scenes/events from multiple fixed-location vision sensors, such as video camcorders, infrared cameras, Time-of-Flight sensors etc, is of broad interest in computer vision society, with many applications including 3D TV, virtual reality, medical surgery, marker-less motion capture, video games, and security surveillance. However, most of the existing multi-view systems are set up in strictly controlled environment, with fixed lighting conditions and simple background views. Many challenges are preventing the technology to the natural environment, including varying lighting, shadows, reflections, motion at the background and visual occlusion etc. In this talk, I address a few aspects to overcome the aforementioned difficulties: (1) occlusion handling, (2) heterogeneous sensor collaboration, and (3) 3D dense motion field estimation, so as to reduce human preparation and manipulation, towards making a robust system as automatic as possible.
Bio: Li Guan is a post-doc researcher in the Computer Vision Lab of Computer Science department, University of North Carolina-Chapel Hill, where he defended his PhD degree in August 2009 and received his MS in 2007. He has been a research assistant (supervised by Professor Marc Pollefeys) at UNC since 2004. His research is primarily about 3D reconstruction problems from varieties of vision sensors using Bayesian inference and discrete optimization. He is also interested in 3D motion analysis & tracking, volume rendering, distributed sensor communication network and GPU computing. Li has published in several top international computer vision conferences and book chapters. He also serves as reviewers for many top computer vision conferences.
Talk

Name: Wolfram Burgard, University of Freiburg
Date: Tuesday, December 15
Time: 3:30 - 4:30pm
Host: Dieter Fox
Title: 3D-Mapping with Mobile Robots
Abstract: Three-dimensional mapping is an essential pre-condition for many robot tasks. To reliably operate, mobile robot must be aware of the three-dimensional structure of their environment, for example, to reliably plan their paths or to perform obstacle avoidance. Additionally, three-dimensional maps are essential for search and manipulation tasks as they allow the robot to plan its actions. In this presentation, I will describe some of the recently developed methods for three-dimensional mapping. This includes compact representations, techniques for solving the simultaneous localization and mapping problem, inference techniques, as well as applications. All methods have been evaluated on mobile platforms and using real data. I will present several experiments demonstrating the advances of the individual techniques.
Bio: Wolfram Burgard studied Computer Science at the University of Dortmund and received his Diploma Degree in 1987. In 1990, he became a member of the University of Bonn where he got his Ph.D.~in 1991. From 1991 to 1999 Wolfram Burgard was a PostDoc at the University of Bonn. Since 1999, Wolfram Burgard is a professor at the University of Freiburg where he heads the Laboratory for Autonomous Intelligent Systems (AIS). His major research interests include mobile robotics, state estimation and control, and artificial intelligence. In his career, Wolfram Burgard has published two books and over 200 papers and articles in outstanding journals and conference proceedings. For his scientific contributions he has received eight best paper awards
from major conferences including the IEEE International Conference on Robotics and Automation (ICRA), IEEE International Conference on Intelligent Robots and Systems (IROS), and the National Conference on Artificial Intelligence (AAAI). Wolfram Burgard is an active member of the IEEE Robotics and Automation Society as well as a life-time member of the American Association of Artificial Intelligence (AAAI). He is Fellow of the European Coordination Committee for Artificial Intelligence (ECCAI) and the Association for the Advancement of Artificial Intelligence (AAAI). In 2009, Wolfram Burgard received the Gottfried Wilhelm Leibniz Prize of the German Research Foundation (DFG), the most prestigious German research award.
Talk

Name: Bryan Russell, Ecole Normale Superieure
Date: Monday, December 14
Time: 10:00 - 11:30am
Host: Xiaofeng Ren
Title: LabelMe: online image annotation and applications
Abstract: Central to the development of computer vision systems is the collection and use of annotated images spanning our visual world. Annotations may include information about the identity, spatial extent, and viewpoint of the objects present in a depicted scene. Such a database is useful for the training and evaluation of computer vision systems. Motivated by the availability of images on the internet, we introduced a web-based annotation tool that allows online users to label objects and their spatial extent in images. To date, we have collected over 500K annotations that span a variety of different scene and object classes. In this talk, I will show the contents of the database, its growth over time, and statistics of its usage. In addition, we use the collected user-provided object annotations to extract the real-world 3D coordinates of images in a variety of scenes. Important for this task is the recovery of geometric information that is implicit in the object labels, such as qualitative relationships between objects (attachment, support, occlusion) and quantitative ones (inferring camera parameters). We show that we are able to obtain high quality 3D information by evaluating the proposed approach on a database obtained with a laser range scanner. Joint work with Antonio Torralba (MIT) and William T. Freeman (MIT)
Bio: Bryan Russell is a postdoctoral fellow in the INRIA Willow team at the Ecole Normale Superieure in Paris, France. His research is in the area of computer vision, with particular interests in object recognition and scene understanding. Bryan received his PhD in 2007 from MIT.
Talk

Name: Daniel Ashbrook, Georgia Tech
Date: Friday, December 4
Time: 10:00 - 11:00am
Host: Jeff Hightower
Title: Situational Impairments and Microinteractions
Abstract: The lines between mobile devices - such as PDAs, mobile phones and cameras - have blurred, and are now almost gone. Similarly, the lines between the resulting technology and we, as human beings, are blurring. The challenge for the field of Computer Science, and in particular Human-Computer Interaction, is to ensure that the technology seamlessly fades into the fabric of our lives, rather than requiring ever more of our attention. One major obstacle preventing technology from becoming a background part, rather than a focus, of our lives, is the usability of the interface while in real-world situations. Many mobile devices borrow from the desktop paradigm, assuming a reasonably quiet, stationary, and controlled environment. Real life, however, isn't so stable, presenting us with many situations that the interfaces to our current mobile devices handle poorly. Rather than treating life as a problem to be solved, we need to make our technology conform to life. In this talk, I will consider three commonly encountered situations that pose difficulties when we interact with devices, discuss how those difficulties may be studied, and present some of my research into helping mobile devices work better in real-world circumstances. Joint work with Antonio Torralba (MIT) and William T. Freeman (MIT)
Bio: Daniel Ashbrook is a Ph.D. candidate in Computer Science in the School of Interactive Computing at the Georgia Institute of Technology. He is a member of the Contextual Computing Group, and his research focuses on human-computer interaction with respect to mobile, wearable and ubiquitous computing. He is particularly interested in novel interactions with on-body devices, and how such interactions can be made fast and subtle.
Talk

Name: Landon Cox, Duke University
Date: Wednesday, October 28
Time: 4:00 - 5:00pm
Host: Jaeyeon Jung
Title: Learning to Trust Again: Building Trust from the Ground Up in Mobile Social Services
Abstract: Mobile social services take advantage of the nearly constant physical proximity of devices to their owners to enable a wide range of new social interactions. In a conventional service, devices send intermittent location updates to a provider, which uses those locations to coordinate interactions among participants. Unfortunately, ensuring users' privacy in these systems is difficult, since service providers could leak or misuse users' sensitive location information and users may wish to remain anonymous to each other (except, of course, when they don't). In this talk, I will briefly describe several projects aimed at understanding alternative, privacy-preserving architectures for mobile social services. In particular, I will focus on the SMILE missed-connections service, which allows strangers who were at the same place, at the same time to anonymously contact each other later on. SMILE establishes trust between strangers by giving them proof that they had a genuine encounter, without relying on trusted infrastructure or a trusted service provider. Paper link: http://www.cs.duke.edu/~lpcox/smile-ccs09.pdf
Bio: Landon Cox is an Assistant Professor of Computer Science at Duke University, and is a recipient of an NSF CAREER award and an IBM Faculty Award. His research has received support from the NSF, AT&T Labs, Amazon, Nokia Research, IBM, and Intel Labs. Landon received his Ph.D. from the University of Michigan in 2005, and his current research interests include operating systems, distributed systems, and mobile computing with a focus on privacy and security.
Talk

Name: Morgan Quigley, Stanford University
Date: Wednesday, October 7
Time: 12:00 - 1:00pm
Host: Josh Smith
Title: Towards Low-Cost Robotic Manipulation in Unconstrained Environments
Abstract: Advances in manipulator design and motion-planning algorithms have resulted in industrial manipulators with astonishing repeatability, precision, and speed. However, the successes of industrial production-line robots have not yet led to widespread deployment of general-purpose robotic manipulators in everyday homes and workplaces. One contributing factor is the high cost of current human-scale manipulators, which makes them economically infeasible in many potential applications, and limits research to only well-equipped laboratories. Although increased production volumes would certainly reduce the cost of robotic manipulators, we propose an alternative approach: a reduction of the mechanical requirements. We note that errors originating in sensors and algorithms for robotic perception are often far greater than the repeatability specifications of current manipulators. Furthermore, manipulation speed is often less critical in unconstrained environments. We propose that low-cost manipulators, combined with modern sensors and passively compliant end-effectors, will be able to perform a useful set of tasks in everyday environments, and will enable more aggressive research and field-testing. In this talk, I will discuss designs and present experiments with an early prototype of these concepts.
Bio: Morgan Quigley is a Ph.D. Candidate in the Artificial Intelligence Lab of the Computer Science Department of Stanford University, advised by Prof. Andrew Ng. His research interests include perception, control, and software systems of robots designed for everyday environments. He also helps lead the development of the Robot Operating System http://www.ros.org, an open-source robot software framework.

Name: Mastooreh Salajegheh, PhD Student U of Massachusetts Amherst
Date: Wednesday, July 22
Time: 4:00 - 5:00pm
Host: Josh Smith
Title: CCCP: Secure remote storage for computational RFIDs
Abstract: Passive RFID tags harvest their operating energy from an interrogating reader, but constant energy shortfalls severely limit their computational and storage capabilities. We propose Cryptographic Computational Continuation Passing (CCCP), a mechanism that amplifies programmable passive RFID tags’ capabilities by exploiting an often overlooked, plentiful resource: low-power radio communication. While radio communication is more energy intensive than flash memory writes in many embedded devices, we show that the reverse is true for passive RFID tags. A tag can use CCCP to checkpoint its computational state to an untrusted reader using less energy than an equivalent flash write, thereby allowing it to devote a greater share of its energy to computation. Security is the major challenge in such remote checkpointing. Using scant and fleeting energy, a tag must enforce confidentiality, authenticity, integrity, and data freshness while communicating with potentially untrustworthy infrastructure. Our contribution synthesizes well known cryptographic and low-power techniques with a novel flash memory storage strategy, resulting in a secure remote storage facility for an emerging class of devices. Our evaluation of CCCP consists of energy measurements of a prototype implementation on the batteryless, MSP430-based WISP platform. Our experiments show that—despite cryptographic overhead— remote checkpointing consumes less energy than checkpointing to flash for data sizes above roughly 64 bytes. CCCP enables secure and flexible remote storage that would otherwise outstrip batteryless RFID tags’ resources. This work will appear at USENIX Security 2009. Paper: http://www.cs.umass.edu/~negin/files/salajegheh-CCCP-usenix09.pdf
Bio: Mastooreh Salajegheh is a PhD student in the Computer Science Department of the University of Massachusetts Amherst. Her research interest lies in analyzing/providing security in the resource constrained devices such as RFIDs, implantable medical devices, wireless sensor nodes and mobile phones. Mastooreh is a member of RFID CUSP (Consortium for Security and Privacy) and works under the supervision of Professor Kevin Fu. She received her Master in information networking from Carnegie Mellon University. http://www.cs.umass.edu/~negin
Talk

Name: Ryan Wistort, MIT
Date: Wednesday, July 8
Time: 4:00 - 5:00pm
Host: Josh Smith
Title: Personal Robots - Why Social Interaction is Important
Abstract: The Personal Robots Group focuses on developing the principles, techniques, and technologies for personal robots. Ongoing research includes the development of socially intelligent robot partners that interact with humans in human-centric terms, work with humans as peers, and learn from people as an apprentice. More recent work investigates the impact of long-term HRI applied to communication, quality of life, health, and educational goals. In this talk, the ongoing research in the Personal Robots Group will be presented in an effort inform the way we think about robots as they move from factories to everyday environments
Bio: Ryan received an undergraduate degree in Electrical Engineering and Embedded Systems form the University of Washington and is currently pursuing a master’s degree at the MIT Media Laboratory in the Personal Robots Group. His passion is exploring the space where software, hardware, mechanics, and design converge. His current research interests include ways in which social robot agents can scaffold creativity and learning for children.
Talk

Name: Shwetak Patel, UW
Date: Wednesday, June 24
Time: 4:00 - 5:00pm
Host: Sunny Consolvo
Title: New Directions for Infrastructure Mediated Sensing
Abstract: The use of sensing systems in the home has the potential to impact various application areas such as chronic care management, aging in place, and sustainability. However, a major challenge remains in identifying and developing truly ubiquitous sensing solutions that address deployment challenges of cost-effectiveness, installation, maintenance, and overall acceptability for a general audience. In the home, the goal of practical ubiquity has led me to develop a new sensing approach, which I call "Infrastructure Mediated Sensing," or IMS. Infrastructure mediation refers to the simple augmentation and probing of existing home infrastructure, such as the electrical power lines, plumbing, or HVAC systems, to sense human activity. In this talk, I will discuss my work in infrastructure and wireless sensing, applications of these technologies, and some new research directions.
Bio: Shwetak N. Patel is an Assistant Professor in the departments of Computer Science and Engineering and Electrical Engineering at the University of Washington. Dr. Patel is also a member of dub. His research interests are in the areas of Human-Computer Interaction, Ubiquitous Computing, and User Interface Software and Technology. He is particularly interested in developing easy-to-deploy sensing technologies and approaches for location and activity recognition applications. Shwetak is also the o-founder of Unsenso, Inc., a demand side energy monitoring solutions provider. Shwetak received his Ph.D. in Computer Science from the Georgia Institute of Technology in 2008 and B.S. in Computer Science in 2003. He was also the Assistant Director of the Aware Home Research Initiative at Georgia Tech.

Name: Yang Wang, Simon Fraser University
Date: Wednesday, June 17
Time: 4:00 - 5:00pm
Host: Xiaofeng Ren
Title: Conservation of Human Attention: Approaches and Prospects
Abstract: One of the grand challenges in computer vision is to enable machines to "see people". The solution to this problem will have tremendous impact on our daily life. The problem also poses many technical challenges for computer vision and machine learning research, due to the overall complexity of the image data and the inherent rich structures in the problems related to ``seeing people". In this talk, I will present my work on two problems in the general area of ``looking at people'', namely, human pose estimation and human action recognition. In particular, I will formulate the solutions to these problems via learning structured models. Firstly, I will present a boosting-style approach for learning structured output models and apply it in human pose estimation. Secondly, I will talk about a max-margin framework for learning structured latent variables models. We apply this framework to the problem of part-based human actions recognition. Thirdly, I will present a bag-of-words model for capturing temporal information of videos for action recognition, which achieves state-of-the art results on several benchmark datasets. I will also highlight a broader range of possible applications of our proposed approaches in computer vision (e.g., object recognition, image segmentation, scene understanding), and possibly other fields (e.g., natural language processing, information fusion) of computer science.
Bio: Yang Wang is currently finishing his PhD in Computing Science at Simon Fraser University, working with Prof. Greg Mori on computer vision and machine learning. He received his M.Sc. from University of Alberta, and his B.Sc. from Harbin Institute of Technology, both in computer science. He worked as a research intern at Microsoft Research Cambridge in summer 2006. His Ph.D. work mainly focuses on high-level recognition problems in computer vision; in particular, human action recognition and human pose estimation. He also works on related machine learning techniques including structured prediction, discriminative learning, graphical models and approximate inference.
Talk

Name: Alyssa Apsel, Cornell University
Date: Thursday, June 11
Time: 10:00 - 11:00am
Host: Joshua Smith
Title: Conservation of Human Attention: Approaches and Prospects
Abstract: As IC applications have multiplied over the past decade, pushing CMOS electronics beyond the PC and into everything from greeting cards to the human body, so have problems associated with nano-scale high performance CMOS. The quest for improved performance, previously masked by the progression of Moore’s law, now calls for renewed creativity and the development of fundamentally new approaches to circuit and architecture design. In this talk, I will consider how the progression of CMOS digital electronics and devices optimized for digital performance has affected mixed signal circuit design. I will offer examples from my research of new approaches to low power RF transceiver and frequency synthesizer design that leverage today’s technology but achieve improved performance per unit power. I will also consider how some of the problems resulting from device scaling such as process variation, noise, and reduced analog performance can be addressed with skillful analog and mixed signal design.
Bio: Alyssa Apsel joined Cornell University in 2002, where she is currently an Associate Professor of Electrical and Computer Engineering. The focus of her research is on power-aware mixed signal circuits and solving the problems that arise in highly scaled CMOS and modern electronic systems. She has authored or coauthored over 65 refereed publications in related fields of RF mixed signal circuit design, interconnect design and planning, photonic integration with VLSI, and circuit design techniques in the presence of variation resulting in five patents and several pending patent applications. She received a best paper award at ASYNC 2006, had a MICRO “Top Picks” paper in 2006, received a college teaching award in 2007, received the National Science Foundation CAREER Award in 2004, and was selected by Technology Review Magazine as one of the Top 100 Young Innovators in 2004. She has also served as an Associate Editor of IEEE Transactions on Circuits and Systems I and II and IEEE Transactions on VLSI
Talk

Name: Scott Hudson, CMU
Date: Wednesday, June 10
Time: 4:00 - 5:00pm
Host: Daniel Avrahami
Title: Conservation of Human Attention: Approaches and Prospects
Abstract: In 1969 Herbert Simon put forward the idea that: “in an information rich world, the scarce resource is [human] attention.” Today this would seem to be increasingly true. This talk will suggest that a number of the important challenges for ubiquitous computing – including goals we describe as “invisible”, “ambient”, “pervasive”, “ubiquitous” or even “calm” – are fruitfully considered in these terms. I will suggest that traction can be gained by focusing on human attention as an organizing paradigm, describe some specific approaches to building systems in this light, and consider where the most fruitful research challenges for future work in this area may lie.
Bio: Scott Hudson is a Professor in the Human-Computer Interaction Institute within the School of Computer Science at Carnegie Mellon University where he directs the HCII PhD program. Elected to the CHI Academy in 2006, he has published extensively on technology-oriented HCI topics. He has regularly served on program committees for the ACM SIGCHI and UIST conferences, and is currently serving as the papers co-chair for the SIGCHI 2010 conference.
Talk

Name: Julie Letchner, UW
Date: Wednesday, March 25
Time: 4:00 - 5:00pm
Host: Matthai Philipose
Title: Access Methods for Markovian Streams
Abstract: In this talk I introduce three novel indexes for answering event (sequence) queries on Markovian streams. Markovian streams are a common class of probabilistic, correlated data sequences inferred from sensor data (e.g. uncertain location sequences inferred from RFID readings, uncertain sentences inferred from raw audio, etc.). Because of the correlations present in these streams, standard query processing methods require sequential examination of an entire stream. This talk presents three access methods that avoid this full stream scan, reducing query time by several orders of magnitude. Results are demonstrated on an uncertain location dataset inferred from real-world RFID traces.
Bio: Julie Letchner is in the final year of her PhD in UW's CSE department. Her work explores the intersection of machine learning and large-scale data management techniques. Julie's work is supported by NSF, NDSEG, ARCS and Google graduate research fellowships. Prior to her graduate studies at UW, Julie earned her B.S. and M.S. from Stanford University with a focus in Artificial Intelligence and Robotics.

Name: Tomas Isdal, UW
Date: Thursday, March 19
Time: 1:00 - 2:00pm
Host: Ken Lafond
Title: Friend-to-Friend data sharing with OneSwarm by Tomas Isdal
Abstract: Peer-to-peer data distribution is a useful primitive for building scalable network services. But, the public sharing typical of currently popular P2P networks exposes user behavior to systematic monitoring, raising privacy concerns that restrict use. In this talk I will present the design and implementation of OneSwarm, a P2P data sharing protocol that uses strong identities and friend-to-friend data transfer to inhibit systematic monitoring of its users. To guide our design, we report measurements of the social network and sharing workload from a popular music web service with more than one million users. Trace playback and measurements of real-world deployments conﬁrm the practicality of our approach. OneSwarm effectively controls overlay overhead, stress, and stretch while restricting sharing to a social graph.
Bio: Tomas Isdal is a graduate student in computer science at the University of Washington. His interests include peer-to-peer and distributed systems, internet measurements, and network security and privacy.
Talk

Name: Jin Wang, UW
Date: Wednesday, Feb 11
Time: 4:00 - 5:00pm
Host: Benjie Limketkai
Title: Bio-inspired design of actuators and sensors based on Ionic Polymer Metal Composites (IPMCs)
Abstract: Ionic Polymer Metal Composites (IPMCs), as a newly-emerged electroactive polymer, have been widely investigated as bending actuators and sensors due to their good performance, such as low actuation voltage, large strain as well as good compatibility in wet environment. In this talk, two typical ionic polymer Nafion and Flemion will be briefly introduced and compared. We illustrate how IPMCs can work as actuators and sensors. Prototype devices done at CIMS will be introduced as well as demos done by other labs. By applying a novel bio-inspired design, a tactile sensor array can be achieved with vectorial force sensing capability. We will also briefly discuss possible applications especially in medical field.
Bio: Ms. Jin Wang, PhD candidate in Materials Science and Engineering Department, UW Ms. Jin Wang is currently a research assistant at Center for Intelligent Materials and Systems, directed by Prof. Minoru Taya at University of Washington. She obtained both her master and bachelor degrees at Tsinghua University, Beijing, China. In summer 2005, she came to UW and started the research on ionic polymer metal composites (IPMCs). Her major research interests include surface electrode on ionic polymers, lifespan improvement for IPMC devices, bio-inspired design of IPMC sensors and packaging issues. Ms. Jin Wang received Global Center of Excellent (G-COE) Fellowship by Japan Society for the Promotion of Science, and stayed at Osaka University under supervision of Prof. Michio Matsumura for research on 3-D nanostructures based on single crystal Si from Dec.2007 to Jan 2008. Ms. Jin Wang is expecting to get her PhD degree in March 2009.
Talk

Name: Margaret (Margie) Morris, Intel (Digital Health Group)
Date: Monday, Feb 2
Time: 3:00 - 4:00pm
Host: Sunny Consolvo
Title: Facilitating emotional self-regulation and preventive cardiology in everyday life: Themes from an exploratory field study of the "Mood Phone"
Abstract: Psychotherapies, particularly those focused on cognitive reappraisal and mindfulness, foster self-awareness and emotional regulation. The ability to modulate reactions to stress can improve immediate wellbeing and possibly reduce emotional risk factors for cardiovascular disease. But these practices are limited in availability: They are affordable by few and generally not available at the moments of greatest stress. In this project, a mobile application (the “Mood Phone”) was created to translate extended clinical dialogues into brief, experiential interfaces that can be used in the flow of daily life. Cognitive reappraisals, breathing and physical relaxation exercises, along with novel self-assessment scales, were designed for a touch screen mobile phone. This mobile application emerged from a research project called Mobile Heart Health that explored the responsiveness of mobile therapies to wireless cardiovascular sensing. The Mood Phone was tested with 53 people in exploratory field trials for periods of one week and with ten individuals seeking stress reduction in a one month study. Participants were prompted for their moods throughout the day and interviewed weekly about their use of the application for stress reduction and their reaction to trends of their mood as it varied with time, social context, and location. Video case studies illustrate the ways in which people applied the phone to increase self-awareness, manage stress and change interpersonal communication.
Bio: Margaret (Margie) Morris is a Senior Researcher in Intel’s Digital Health Group. She is a clinical psychologist who explores emerging technologies, particularly mobile devices, for enhancing emotional and physical wellbeing. She led the Mobile Heart Health project and has recently completed studies on chronic disease, social engagement, and cognitive aging. Prior to joining Intel, she studied technology adoption at Sapient Corporation. She completed her Ph.D. in Clinical Psychology with a minor in Behavioral Neuroscience at the University of New Mexico, her clinical internship at the San Francisco VA Medical Center, and her postdoctoral fellowship at Stanford University. She has a B.A. in English from Haverford College.

Name: Andrew Ko, UW
Date: Wednesday, Jan 14
Time: 4:00 - 5:00pm
Host: Sunny Consolvo
Title: Debugging Reinvented: Asking and Answering Why and Why Not Questions about Program Behavior
Abstract: Most software undergoes a brief period of rapid development, followed by a much longer period of maintenance and adaptation. As a result, software developers spend most of their time exploring a system's underlying source code to determine the parts of the system that are relevant to their tasks. Because these parts are often distributed throughout a system's modules, and because they can interact in complex and unpredictable ways, this process of understanding a program's execution can be extremely difficult. The primary cause of this difficulty is that developers must answer their questions about a system's behavior by guessing. For example, a developer wondering, "Why didn't this button do anything after I pressed it?" must form an answer such as "Maybe because its event handler wasn't called" and then use breakpoint debuggers, print statements, and other low-level tools to verify the explanation. Not only is this process poorly supported by current tools, but worse yet, there are many potential explanations for a system's behavior, so developers rarely formulate a valid explanation on the first attempt. To address this problem, I present a new kind of program understanding tool called a Whyline, which allows a developer to select "why did" and "why didn't" questions directly about the symptoms of a system's behavior. In response, the Whyline determines which parts of the system and its execution are related to the symptom in question, while also identifying false assumptions the developer might have about what occurred during the execution of the program. By using this approach, developers need not guess about potential causes of program behavior: they simply point to some perceptible feature of the faulty behavior and the system identifies the relevant code. Early prototypes of the Whyline for a simplified educational programming language reduced debugging time by a factor of 8. I have since generalized the Whyline to support Java programs with textual and graphical output, inventing several new incremental algorithms to identify program-specific output, derive output-relevant questions, and answer questions about a variety of output. In a comparison of the Whyline to conventional debugging tools, developers with the Whyline found bugs with twice the success, twice as fast.
Bio: Andrew Ko is an Assistant Professor at The Information School at theUniversity of Washington. His research interests include social and cognitive factors in software engineering, end user software engineering, user interface software and technology, and programming language design. He has published articles in all of these areas, receiving best paper awards at top conferences such as the International Conference on Software Engineering (ICSE) and the ACM Conference on Human Factors in Computing (CHI), as well as extensive press on the Whyline, a novel debugging tool that supports questions about program output. In 2004, he was also awarded both NSF and NDSEG research fellowships in support of his Ph.D. research. He received his Ph.D at the Human-Computer Interaction Institute at Carnegie Mellon University, advised by Brad Myers. He received Honors BS degrees in Computer Science and Psychology from Oregon State University in 2002.
Talk

Name: Alexander Mamishev and Gabriel Rowe, UW
Date: Wednesday, Dec 3
Time: 4:00 - 5:00pm
Host: Josh Smith
Title: Design and Applications of Multifunctional Conformable Sensor Arrays
Abstract: The field of prosthetics is often characterized by media coverage of many unfortunate young war veterans wearing microprocessor controlled prosthetic knees or research breakthroughs in neurointegrated prosthetic arms. However, the largest constituent of amputees are those that lost their limbs not in a war-zone, but rather in the course of diabetes. Approximately 10 million people in the US have diabetes, and this number is likely to grow. Diabetic amputees nearly all have a prosthesis that replaces their foot and a portion of their leg below the knee. No matter the type of amputee, the interface between the human and the prosthesis is a source of constant discomfort and dissatisfaction for most amputees. This human-machine-interface is non-optimal at present and many researchers are trying to change this. A "below-knee" prosthesis consists of a soft rubber liner and hard prosthetic socket that attach to a prosthetic leg and foot. This presentation will focus on technologies that have been used within the prosthetic socket to attempt to measure pressure, shear stresses, and temperature in order to attempt to better understand the prosthesis-amputee interface. Flexible pressure sensors, strain gages, small thermistors, and custom socket designs have been created to attempt to measure many parameters within the socket to improve it and learn how the body interacts with directly coupled mechanical devices. If one desires to look forward into the future of human machine interfacing where soldiers and construction workers wear exoskeletons, and the elderly are assisted by robotic transportation devices, the best way to study these problems is to look at the research going into solving the problems of amputees today.
Bio: Alexander V. Mamishev received an equivalent of B.S.E.E. degree from Kiev Polytechnic Institute, Ukraine in 1992, M.S.E.E from Texas A&M University in 1994, and Ph.D. in Electrical Engineering and Computer Science from MIT in 1999, with a minor in Technology Management from Harvard Business School and MIT Sloan School of Management. Currently, he is an Associate Professor, Director of Sensors, Energy, and Automation Laboratory (SEAL), and Director of Electrical Energy Industrial Consortium (EEIC) in the Department of Electrical Engineering, University of Washington, Seattle . Prof. Mamishev is an author of about 100 journal and conference papers, three book chapters, and two patents. His research interests include sensor design and integration, robotics, and energy technology applications. He serves as an Associate Editor for the IEEE Transactions on Dielectrics and Electrical Insulation and a reviewer for several journals and conferences. He is a recipient of the NSF CAREER Award, the IEEE Outstanding Branch Advisor Award, and the UW EE Outstanding Research Advisor Award.
Bio: Gabe Rowe received his BSEE in 2003 and MSEE in 2005 both from the University of Washington. He is currently a PhD candidate in Electrical Engineering at the University of Washington. He has been working in the Center of Excellence in Limb Loss Prevention and Prosthetics Engineering at the Veteran's Affairs Medical Center in Seattle since 2005 toward the goal of creating capacitive sensor arrays that conform to the skin surface with a goal of improving socket fit and amputee thermal comfort. He is author or co-author of 10 journal and conference papers. He has also been working with Cypress Semiconductor since 2007 as an applications engineer in the capacitive touchscreen group.

Name: Garrett Cobarr, Swissopolis GmbH
Date: Wednesday, Nov 19
Time: 4:00 - 5:00pm
Host: Beverly Harrison
Title: The Evolution of Virtual Worlds into the 3D Web
Abstract: A brief talk about technologies and social indicators that have led to the current state of virtual worlds. A slightly longer discussion of the current state of virtual worlds and the balkanization of the metaverse. The main point: where is this going? Where should it go? The emergence of the 3D web and need for more than the internet. An explanation of the difference between presentation, interactivity and interaction. The internet was about machines and individual points of broadcast, the metaverse is about the reduction of entropy in human interaction and the need for new mechanisms to support those trends. The metaverse is not a game or a replacement for reality, it is an amplification and augmentation of reality. This effort will require more than just the development of rendering and network systems but peer ascension, Rule of Law mechanisms, economic transactional systems, collaboration models and social and political interaction methodologies.
Bio: BFA in Painting and Drawing, minor in Astronomy. Worked as designer and illustrator for many years. Started working in experimental computer music in early 1980's and by the end of that pursuit in 1989 had received grants from Seattle, King County and Washington State Arts Commissions and the National Endowment for the Arts. Composer in residence for Pacific Science Center. Performed and interviewed on radio 3 times: KRAB once, KUOW twice. Composition was focused on multiple synthesizers controlled by computer played through a circle of speakers around the audience. Increasing work with choreographers in theaters led to greater interest and work with automation via MIDI using an early multimedia authoring tool, Hypercard. Career focus shift. In 1990, I founded my company Lingua Media Group. Design and production of user experience, interface, motion graphics and interactive multimedia with a special focus on 3D. Prototyping, interface and simulations for Microsoft, Intel, The Casey Family Program, Microvision and almost all design and advertising firms located in Seattle. Lingua Media Group has worked with over 200 startups. A lecture to the Graphic Arts Guild in 1996 led to an offer to teach. The next 6 years I taught at Shoreline, Seattle Central, North Seattle Community Colleges, Seattle Art Institute and the University of Washington, where I was nominated for a teacher of the year award. In the late 1990's I worked on and consulted for a number of internet web sites but moved to design of intranets and extranets. Many of those projects were with firms like KPMG, Anderson and Deloitte & Touche. Entered Second Life in March 2006. Bought an island region, Lingua Franca in late 2006 and founded the Island Region Sim Owners group, about a hundred members with about 3800 island sims. Met and was asked to design a virtual world project for a Swiss firm in May of 2007. Project incorporated as Swissopolis GmbH in late 2007 and became a major partner in charge of technology, design and strategy. Became Chief Executive Officer in March 2008. We are working on the opening of our project Second Life by the end of 2008. Started the DaVirtua initiative to explore the emergence of the 3D web.
Talk

Name: Pavel Nikitin, Intermec Technologies
Date: Wednesday, Nov 12
Time: 4:00 - 5:00pm
Host: Josh Smith
Title: Physical layers of non-traditional wireless systems.
Abstract: In this talk, physical layers of two non-traditional wireless systems, HVAC duct network and passive RFID, are discussed. HVAC duct system of a building is a network of interconnected hollow metal pipes which can behave as waveguides and be used for networking. We discuss antenna and channel properties of such communication system, including mode excitation, attenuation, and dispersion, and their influence on data throughput. Experimental measurements on real HVAC ducts are also included. RFID is a wireless technology based on modulated backscatter. We present an RF-centric overview of UHF RFID systems and discuss forward and reverse links using simple analytical channel model. The model is supported with experimental measurements and can be used for analysis of various systems. Several practical RFID antenna design examples and various applications of RFID technology are also described.
Bio: Dr. Pavel Nikitin is a Lead Engineer at Intermec Technologies, Everett, WA where he designs and develops RFID tags and systems. He is also an Affiliate Assistant Professor in Electrical Engineering Department at University of Washington. He received Ph.D. in Electrical and Computer Engineering in 2002 from Carnegie Mellon University and worked at Ansoft and IBM. He is the author of approximately 50 technical publications and 20 patent applications.
Talk

Name: Ramon Caceres, AT&T Labs
Date: Thursday, Oct 30
Time: 11:00 - 12:00pm
Host: David Wetherall
Title: Trustworthy and Personalized Computing on Public Kiosks
Abstract: Many people desire ubiquitous access to their personal computing environments. We present a system in which a user leverages a personal mobile device to establish trust in a public computing device, or kiosk, prior to resuming her environment on the kiosk. We have designed a protocol by which the mobile device determines the identity and integrity of all software loaded on the kiosk, in order to inform the user whether the kiosk is trustworthy. Our system exploits emerging hardware security technologies, namely the Trusted Platform Module and support in x86 processors for establishing a dynamic root of trust. We have demonstrated the viability of our approach by implementing and evaluating our system on commodity hardware. This is joint work with Scott Garriss (Google), Stefan Berger (IBM), Reiner Sailer (IBM), Leendert van Doorn (AMD), and Xiaolan Zhang (IBM).
Bio: Ramon Caceres is a Lead Member of Technical Staff at AT&T Labs. His research interests include mobile/pervasive/ubiquitous computing, wireless networking, virtualization, and security. He has previously been a Research Staff Member at IBM Research and Chief Scientist of Vindigo, an award-winning provider of location-based services for mobile devices. He holds a Ph.D. from the University of California at Berkeley and is an ACM Distinguished Scientist.
Talk

Name: Patrick Baudisch, Microsoft Research
Date: Wednesday, Oct 15
Time: 4:00 - 5:00pm
Host: Sunny Consolvo
Title: Back-of-Device Interaction Allows Creating Very Small Touch Devices
Abstract: In this talk, I will discuss how to add pointing input capabilities to very small screen devices. On first sight, touchscreens seem to allow for particular compactness, because they integrate input and screen into the same physical space. The opposite is true, however, because the user's fingers occlude contents and prevent precision. I argue that the key to touch-enabling very small devices is to use touch on the device backside. In order to study this, we have created a 2.4" prototype device; we simulate screens smaller than that by masking the screen. I present a user study in which participants completed a pointing task successfully across display sizes when using a back-of device interface. The touchscreen-based control condition (enhanced with the shift technique), in contrast, failed for screen diagonals below 1inch. I present four form factor concepts based on back-of-device interaction and provide design guidelines extracted from a second user study.
Bio: Patrick Baudisch is a research scientist in the field of human-computer interaction at the Adaptive Systems and Interaction Research Group at Microsoft Research, as well as an Affiliate Assistant Professor in Computer Science at the University of Washington. His research focus is on interaction with small screen devices, which evolved from a longer series of research projects on interaction with large displays. While at Xerox PARC, Baudisch created focus plus context screens. While at Fraunhofer-IPSI and during his stay as a guest researcher at the GroupLens project at the University on Minnesota, Baudisch worked on user interfaces for information filtering systems. He holds a PhD in Computer Science from Darmstadt University of Technology, Germany. More info: http://www.patrickbaudisch.com
Talk

Name: Saul Griffith, Makani Power
Date: Wednesday, October 8
Time: 11:00 - 12:00pm
Host: Josh Smith
Title: Renewable energy, etc.
Abstract: Saul Griffith will talk about global energy limit cases and calculating your own personal energy use (www.wattzon.org), high altitude wind power
(www.makanipower.com), human based power (www.potenco.com), and kite powered hydrofoils(www.kiteboat.com), maybe a little children's science
education (www.howtoons.com) just for good measure
Bio: Saul Griffith is President & Chief Scientist of Makani Power and a 2007 MacArthur foundation fellow. Dr. Saul Griffith has multiple degrees in materials science and mechanical engineering and completed his PhD in Programmable Assembly and Self Replicating machines at MIT. He is the co-founder of numerous companies including: Low Cost Eyeglasses, Squid Labs, Potenco, Instructables.com, HowToons and Makani Power. Saul has been awarded numerous awards for invention including the National Inventors Hall of Fame, Collegiate Inventor's award, and the Lemelson-MIT Student prize. A large focus of Saul's research efforts are in minimum and constrained energy surfaces for novel manufacturing techniques and other applications. Saul holds multiple patents and patents pending in textiles, optics, nanotechnology, and energy production. Saul co-authors children's comic books called "HowToons" about building your own science and engineering gadgets with Nick Dragotta and Joost Bonsen. Saul is a technical advisor to Make magazine and Popular Mechanics. Saul is a columnist and contributor to Make and Craft magazines.

Name: Donald S. Gardner, Intel Corp
Date: Thursday, Sep 18
Time: 2:00 - 3:00pm
Host: Josh Smith
Title: Integrated On-Chip Inductors Using Magnetic Material
Abstract: On-chip inductors with magnetic material are integrated into both advanced 130 and 90 nm CMOS processes. The inductors use aluminum or thick copper metallization and amorphous CoZrTa magnetic material. Increases in inductance of up to 30 times corresponding to an inductance density of up to 1,700 nH/mm2 were obtained, significantly greater than prior values for on-chip inductors with magnetic material. In comparison, air-core spiral inductors can achieve inductance densities of up to about 100 nH/mm2. With such improvements, the effects of eddy currents, skin effect, and proximity effect become clearly visible at higher frequencies. The CoZrTa was chosen for its good combination of high permeability, good high-temperature stability (>250 °C), high saturation magnetization, low magnetostriction, high resistivity, minimal hysteretic loss, and compatibility with silicon technology. The CoZrTa alloy can operate at frequencies up to 9.8 GHz, but trade-offs exist between frequency, inductance, and quality factor. The effects of increasing the magnetic thickness on the permeability were measured and modeled including skin depth effects, eddy current dampening, and the effects of the demagnetization field. The inductors use magnetic vias and elongated structures to take advantage of the uniaxial magnetic anisotropy. Techniques are presented to extract a sheet inductance and examine the effects of magnetic vias (vias that allow complete closure in the magnetic flux) on the inductors. Comparisons of measurements of different via width and of structures with versus without laminations demonstrate the effectiveness of this technique with thin cobalt oxide. Comparing inductors with maximum Q-factors at different frequencies was accomplished by plotting the inductance over ac resistance time constant (L/Rac) versus frequency, then including contours representing constant quality-factor values. Simulations of magnetic flux density and eddy current densities and analytical models were used to gain a good understanding of the effects of laminations. The inductors with thick copper and thicker magnetic films were successfully demonstrated to have L/Rac time constants about 20× higher than earlier aluminum-based inductors with resistances as low as 0.04 W and quality factors of up to 8 at frequencies as low as 40 MHz.
Bio: Donald Gardner has been with Intel Corporation since 1991 and is currently a Principal Engineer in Intel Research. He is also is a visiting scientist at Stanford University and a senior member of the IEEE. He received his PhD in Electrical Engineering from Stanford University. Donald is the inventor or co-inventor of 58 patents including for inductors using high-frequency magnetic materials, reflow of copper metallization, layered aluminum metal for interconnections, and embedded ground planes. He has received Intel’s highest technical award (Intel Achievement Award) for fundamentally changing technologies by incorporating inductors with magnetic material on CMOS. Donald invented a copper process and used it to fabricate the first working chip with copper-based interconnections at Intel, then published papers on copper size effects that has been referred to as the first study that showed surface scattering and grain size to be an interconnect scaling issue. He also invented an Al alloy/Ti metallization for interconnections as part of his PhD thesis studies that was later widely used in microchips. Donald has published and presented over 140 electrical engineering, materials science and computer science papers in journals and conferences including several invited presentations. He has received three Best Paper and Poster awards at international conferences and over 500 authors have cited his publications. Donald has had appointments as a visiting research scientist at Hitachi Research Labs in Japan and as an instructor at Stanford University. He enjoys bringing new life to old technologies by blending them with different technologies or recent science and new materials. His current interests include magnetic materials for high-frequency inductors, silicon-based optoelectronic devices, nanostructure design and devices, and process integration.
Talk

Name: Krzysztof Gajos, Microsoft Research
Date: Weds, Sep 17
Time: 4:00 - 5:00pm
Host: Daniel Avrahami
Title: Automatically Generating Personalized User Interfaces
Abstract: User Interfaces delivered with today's software are usually created in a one-size-fits-all manner, making implicit assumptions about the needs, abilities, and preferences of the "average user" and the characteristics of the "average device." I argue that personalized user interfaces, which are adapted to a person's devices, tasks, preferences, and abilities, can improve user satisfaction and performance. I have developed three concrete systems:
-- SUPPLE, which uses decision-theoretic optimization to
automatically generate user interfaces adapted to a person's device
and long-term usage patterns;
-- ARNAULD, which allows optimization-based systems to be adapted to
users' preferences; and
-- Ability Modeler and an extension of SUPPLE that first performs a
one-time assessment of a person's motor abilities and then
automatically generates user interfaces predicted to be the fastest to
use for that user.
My experiments show that these automatically generated, personalized user interfaces significantly improve speed, accuracy, and satisfaction for users with motor impairments compared to manufacturers' default interfaces. I also provide the first characterization of the design space of adaptive graphical user interfaces, and demonstrate how such interfaces can significantly improve the quality and efficiency of daily interactions for typical users. More information about Supple at: http://www.cs.washington.edu/ai/supple/
Bio: Krzysztof Gajos received his B.Sc. and M.Eng. degrees in Computer Science from MIT in 1999 and 2000, respectively. For the next two years, he was a research scientist at the MIT Artificial Intelligence Laboratory, where he managed The Intelligent Room Project. From 2002, he attended the University of Washington, where he worked with Daniel Weld and Jacob Wobbrock. He was a recipient of a Microsoft Graduate Research Fellowship and he has also been a visiting faculty member at the Ashesi University in Ghana, where he designed and taught an introductory course in artificial intelligence. In 2003, he received M.Sc. and in 2008 a Ph.D., both in Computer Science at the University of Washington. Currently he is a postdoctoral researcher at Microsoft Research and in the fall of 2009 he will join the Computer Science faculty at Harvard University. Web page: http://www.cs.washington.edu/homes/kgajos/
Talk

Name: Randy Wang
Date: Weds, Aug 27
Time: 4:00 - 5:00pm
Host: David Wetherall
Title: The Digital StudyHall
Abstract: Digital StudyHall (DSH) seeks to improve education for poor children in slum and rural schools in India. In a nutshell, think of its technical approach as the educational equivalent of YouTube + Netflix + Kazaa.
We digitally record live classes by the best grassroots teachers, transmit them on the "Postmanet" (effected by DVDs sent in the postal system), collect them in a large distributed database, and distribute them on DVDs to poor rural and slum schools. Education experts and teachers use the system to explore pedagogical approaches involving local teachers actively "mediating" the video lessons. By harvesting a "viral phenomenon" of community participation, DSH aims to help train teachers and deliver quality instruction to underprivileged children. The project is a collaboration between computer scientists and education experts.
Talk

Name: Keith Edwards, Georgia Tech College
Date: Thurs, Aug 7
Time: 11:00 - 12:30pm
Host: Jaeyeon Jung
Title: Human-Centered Networking Research at Georgia Tech
Abstract: Networking is moving out of professionally managed environments and into homes, public spaces and other "end user-managed" settings. However, despite the rapid update of networks in such settings, the user experience of networking presents severe challenges for deployment, management, and use. In the US, for example, home networking equipment is the most returned consumer electronics item, largely because of complexity and user experience issues. While networking research has long focused on improving performance, scalability, and security, there has been much less focus on how we create networking technologies that address human-centric concerns, such as understandability, usability, installability, maintainability, and so forth. In this talk I'll describe a research program I lead at Georgia Tech called Human-Centered Networking, aimed at bringing such a human perspective to networking research. I will present an overview of a number of different projects under this program, ranging from empirical studies of the human experience of networking, to design work focused on new visual interfaces and physical form factors for networking equipment, to architectural work aimed at providing a better networking user experience "from the ground up."
Bio: W. Keith Edwards is an Associate Professor in the School of Interactive Computing in Georgia Tech's College of Computing. His research focuses on bringing human perspectives to core computing concerns such as networking and information security, to ensure that these technologies are not only more usable, but also more useful and understandable. Prior to joining the faculty at Georgia Tech he was Principal Scientist at Xerox PARC, where he managed the Ubiquitous Computing group and contributed toward the commercialization of several key technologies. He is the author of two books on Sun's Jini distributed computing framework, over 60 refereed research papers, and almost four dozen US and International patents either granted or pending.
Talk

Name: Jitu Padhye, Microsoft Research
Date: Weds, Aug 6
Time: 4:00 - 5:00pm
Host: Anmol Sheth
Title: Designing High Performance Enterprise Wi-Fi Networks
Abstract: Use of mobile computing devices such as laptops, PDAs, and Wi-Fi enabled phones is increasing in the workplace. As the usage of corporate 802.11 wireless networks (WLANs) grows, network performance is becoming a significant concern. We have built DenseAP, a novel system for improving the performance of enterprise WLANs using a dense deployment of access points (APs). In sharp contrast with wired networks, one cannot increase the capacity of a WLAN by simply deploying more equipment (APs). To increase capacity, the APs must be assigned appropriate channels and the clients must make intelligent decisions about which AP to associate with. Furthermore, the decisions about channel assignment, and associations must be based on a global view of the entire WLAN, rather than the local viewpoint of an individual client or AP. Given the diversity of Wi-Fi devices in use today, another constraint on the design of DenseAP is that it must not require any modification to Wi-Fi clients. In this talk, we show how the DenseAP system addresses these challenges, and provides signify.
Bio: Jitendra Padhye received his PhD in 2000. After a two-year stint at ACIRI (now ICIR), he joined Microsoft Research in 2002, and has been there ever since. His recent research focus has been on wireless networks.
Talk

Name: Brian Kopell, Medical College of Wisconsin
Date: Wed, July 16, 2008
Time: 4:00 - 5:00pm
Host: Ben Greenstein
Title: What's the Problem with Neuromodulation?
Abstract: Neuromodulation is a rapidly emerging field in clinical neuroscience. The two chief techniques of brain stimulation, deep brain stimulation, and cortical stimulation will be defined and discussed. An overview of functional imaging and electrophysiological techniques will be presented to give a context of the tools available to advance the technology of Neuromodulation. Finally a brief discussion of the need for closed-loop stimulation will be presented.
Bio: Dr. Kopell is currently the Assistant Professor of Neurosurgery at the Medical College of Wisconsin, where he specializes in the neuromodulation and functional neurosurgical treatment of various functional neurological maladies such as movement, pain and psychiatric disorders. Brian holds an M.D. from the New York University School of Medicine where he also completed his residency. Brian's fellowships were completed at the University Hospital in Zurich, Cleveland Clinic Foundation and University Hospital in Ankara. Brian's special research interests include new applications of neurostimulation technologies and functional imaging. He is also VP of Medical Affairs for Northstar Neuroscience in Seattle, WA.

Name: Emma Brunskill, MIT
Date: Wed, June 25, 2008
Time: 4:00 - 5:00pm
Host: Ali Rahimi
Title: Learning in continuous-valued domains with noisy offset dynamics
Abstract: Many interesting artificial intelligence planning problems involve continuous-valued state spaces and stochastic, switching dynamics, such as autonomous traversal of varying terrain. In this talk I'll describe a reinforcement learning algorithm for learning and acting in continuous-valued domains with switching noisy offset dynamics. This approach automatically trades off the value of taking an action to better estimate the world dynamics (exploration) versus taking the best action given the current estimate of the dynamics model (exploitation). I'll show that in certain environments the algorithm will perform close to optimally on all but a number of samples that scales polynomially with the state-space dimension. I'll also report the results of an experiment in which a robotic car drives over varying terrain: these results suggest that our dynamics representation can adequately capture real-world dynamics, and that our algorithm can be used to efficiently solve such problems. This is joint work with Bethany Leffler, Lihong Li, Michael Littman and Nicholas Roy.
Bio: Emma Brunskill is a doctoral student in Computer Science at the Massachusetts Institute of Technology. She received a B.S. in Computer Engineering and Physics from the University of Washington and a M.Sc. in Neuroscience from Oxford University as a Rhodes scholar. Her research interests span machine learning, robotics, and the role information communication technologies can play in international development.
Talk

Name: Michael Beigl, Technische Universität Braunschweig
Date: Mon, June 9, 2008
Time: 4:00 - 5:00pm
Host: Anthony LaMarca
Title: Collaboration and Integration in Ubiquitous Computing
Abstract: This talk will present the most recent results in research and development of collaboration-based Ubicomp, and its integration in business and Internet enabled processes. Collaboration, as a main design principle, is e.g. used for exploring low-level networking, namely the Distributed Jam Signalling and Energy Shift Keying concepts. Integration is e.g. used to couple business processes via Internet based systems. This talk will include a presentation of the latest advances in Particle/uPart small wireless sensor system, and Examples of industry-driven research, including hazardous environments, retail-stores and home automation projects.
Bio: Michael Beigl is professor for Ubiquitous and Distributed systems at the Carl-Friedrich Gauss Faculty of the Technische Universität Braunschweig. Before this he was Research Director of TecO, University of Karlsruhe and guest professer at Keio University in 2005. He obtained both his MSc (1995) and PhD (Dr.-Ing)(2000) from University of Karlsruhe. His research interests cover wireless sensor systems, mobile and ubiquitous networks, distribution of Ubicomp enabled information via Internet, location models and systems, novel sensor technology and context awareness.
Talk

Name: Desney Tan of Microsoft Research
Date: Wed, May 21, 2008
Time: 4:00 - 5:00pm
Host: Jaeyeon Jung
Title: Cyberware Engineering: Interfacing Directly with Human Physiological Signals
Abstract: One of the things that distinguishes human beings from other animals is the degree to which we fill our environments with various technologies in order to augment ourselves both physically and cognitively. In fact, we have become human-technology symbionts, quite ineffective at functioning without our various augmentations. In our work, we embrace the notion of human augmentation and propose that there is large untapped potential in interfacing directly with the human body and decoding physiological signals. Specifically, I will describe some of our recent work measuring cognitive load and categorizing images with brain-computer interfaces as well as creating a novel input modality with muscle-computer interfaces. I will also present some of the new projects we are working on in the medical sensing and healthcare domains.
Bio: Desney Tan is a Researcher in the Human Centered Computing research area at Microsoft Research, where he manages the Computational User Experiences (CUE) group. He also holds an affiliate faculty appointment in the Department of Computer Science and Engineering at the University of Washington. Desney's research interests include Human-Computer Interaction and Physiological Computing, specifically Brain-Computer Interfaces. However, he is a somewhat schizophrenic researcher and has worked on projects in domains such as virtual and augmented reality, large and multiple display interfaces, handwriting recognition, as well as adaptive interfaces. Desney received his Bachelor of Science in Computer Engineering from the University of Notre Dame in 1996, after which he spent a couple of years building bridges and blowing things up in the Singapore Armed Forces. He later returned to Carnegie Mellon University, where he worked with Randy Pausch and earned his PhD in Computer Science in 2004. In 2007, he was honored as one of MIT Technology Review's Young Innovators Under 35 for his work on brain-computer interfaces.
Talk

Name: Gonzalo Ramos, Live Labs
Date: Wed, May 14, 2008
Time: 4:00 - 5:00pm
Host: Beverly Harrison
Title: Video Browsing by Direct Manipulation
Abstract: Time is the dominating dimension we use to experience and navigate through video content, yet sometimes thinking about a video in terms of time is not ideal, e.g., when our focus is locked on the objects or scene captured within. In this presentation, I will talk and demonstrate a novel method for browsing videos by directly dragging their content. This browsing method not only makes space a video's dominant dimension, but also brings the benefits of direct manipulation to an activity typically mediated by indirect widgets. I will elaborate as to how we support this new type of interactivity by: 1) automatically extracting motion data from videos; and 2) introducing an interaction technique called relative flow dragging that lets users control video playback by moving objects of interest along their visual trajectory. I will demonstrate a video player that implements browsing by direct manipulation and I will share the results of a study showing that this novel browsing method can out-perform the traditional seeker bar in video browsing tasks that focus on visual content rather than time.
Bio: Gonzalo Ramos received his Honors Bachelors in Computer Science from the University of Buenos Aires where he worked on image compression and wavelets. He later obtained my M.Sc. in Computer Science at the University of Toronto, Canada where he focused on numerical analysis and scientific visualization issues. During his graduate studies, he interned trice at Microsoft Research and was later awarded a Microsoft Research Fellowship. Gonzalo completed his doctoral studies in Computer Science at the University of Toronto where he worked with Professor Ravin Balakrishnan at the Dynamic Graphics Project Lab.
Talk

Name: Ashish Khisti, MIT
Date: Tue, May 13, 2008
Time: 10:00 - 11:30am
Host:Ben Greenstein
Title: Multi-layer architectures for secure communication: information theoretic perspectives
Abstract: In the traditional network hierarchy, reliability and security are handled in different protocol layers. In particular, information is encrypted at the application layer, while lower layers provide an error-free transmission link. Likewise compression is also addressed separately. However, many emerging applications such as wireless ad hoc networks, sensor networks and pay TV systems are vulnerable to new attacks that are not addressed by such separation. In this talk, I will present new architectures in which encryption and source/channel coding are performed jointly, and analyze them within an information theoretic framework. Among other results, we will develop 1) fundamental limits and insights into the role of multiple antennas for protecting confidentiality of information; and 2) source coding techniques for secret key generation and their application to privacy-preserving biometric systems. As will be apparent, good solutions to such problems bring together techniques not only from information theory, but from convex optimization, random matrix theory, signal processing, and graphical models as well. As time permits some recent extensions to joint source and channel coding problems with secrecy constrains will also be discussed.
Bio: Ashish Khisti received the B.A.Sc degree in Engineering Sciences (Electrical Engineering option) from University of Toronto in 2002 and S.M degree in Electrical Engineering and Computer Sciences from Massachusetts Institute of Technology (MIT) in 2004, where he is currently a PhD candidate. His research interests are in the area of information theory and its applications to wireless and multimedia systems. He is a recipient of the Harold L. Hazen Teaching Award and the Joseph Levin Masterworks award from the EECS department at MIT. He is also a recipient of the Hewlett-Packard PhD fellowship, NSERC fellowship for post-graduate studies, and the Lucent global science scholar award. He has been a visiting student at EPFL-Lausanne, ETH-Zurich, and HP Labs, and a summer intern at Mitsubishi Electrical Research Labs (MERL).
Talk

Name: Xiaofeng Ren, Toyota Technological Institute at Chicago
Date: Thur, May 8, 2008
Time: 10:00 - 11:30am
Host: Ali Rahimi
Title: Image and Video Parsing: a Gestalt Approach
Abstract: The grand goal of computer vision is to parse and label every perceptual structure in images. Such a complete understanding requires the use of a wide range of visual cues and the incorporation of associated processes at all levels of abstraction. I have taken an integrated approach to vision with a focus on mid-level processing, including contour/region grouping and figure/ground organization, a crucial part of visual perception that bridges together low-level signals (e.g. edges and texture) and high-level knowledge (e.g. object shape). In this talk I will introduce a compact mid-level image representation using piecewise straight approximation of contours and the constrained Delaunay triangulation (CDT). On top of the CDT graph I will develop a unified probabilistic framework for mid-level vision, using conditional random fields (CRF) to enforce consistencies at junctions. For the first time mid-level vision is shown to be both feasible and useful, through quantitative evaluations on large human-annotated datasets. I will also demonstrate that mid-level representation and processing can apply to, and greatly facilitate, many visual tasks such as tracking objects, segmenting objects from background, and recognizing objects in both still images and videos.
Bio: Xiaofeng Ren received his B.S. from Zhejiang University, his M.S. from Stanford University, and his Ph.D. from the University of California at Berkeley in 2006. He is currently a research assistant professor at the Toyota Technological Institute at Chicago. His research interests lie broadly in the areas of computer vision. His recent work focuses on probabilistic modeling of mid-level vision and its applications in parsing images and video.
Talk

Name: Andreas Krause, CMU
Date: Tue, May 6, 2008
Time: 10:00 - 11:30am
Host: Ali Rahimi
Title: Optimizing Sensing from Water to the Web
Abstract: Where should we place sensors to quickly detect contaminations in drinking water distribution networks? Which blogs should we read to learn about the biggest stories on the web? These problems share a fundamental challenge: How can we obtain the most useful information about the state of the world, at minimum cost? Such sensing, or active learning, problems are typically NP-hard, and were commonly addressed using heuristics without theoretical guarantees about the solution quality. In this talk, I will present algorithms which efficiently find provably near-optimal solutions to large, complex sensing problems. Our algorithms exploit submodularity, an intuitive notion of diminishing returns, common to many sensing problems; the more sensors we have already deployed, the less we learn by placing another sensor. To quantify the uncertainty in our predictions, we use probabilistic models, such as Gaussian Processes. In addition to identifying the most informative sensing locations, our algorithms can handle more challenging settings, where sensors need to be able to reliably communicate over lossy links, where mobile robots are used for collecting data or where solutions need to be robust against adversaries and sensor failures. I will also present results applying our algorithms to several real-world sensing tasks, including environmental monitoring using robotic sensors, activity recognition using a built sensing chair, deciding which blogs to read on the web, and a sensor placement competition.
Bio: Andreas Krause is a Ph.D. Candidate at the Computer Science Department of Carnegie Mellon University. He is a recipient of a Microsoft Research Graduate Fellowship, and his research on sensor placement and information acquisition received awards at several conferences (KDD '07, IPSN '06, ICML '05 and UAI '05). He obtained his Diploma in Computer Science and Mathematics from the Technische Universitat Munchen, where his research received the NRW Undergraduate Science Award.
Talk

Name: Jim Prager, UW
Date: Thur, May 1, 2008
Time: 10:00 - 11:30am
Host: Josh Smith
Title: Experimental Investigation of Plasma Downstream of a High Power Helicon Thruster
Abstract: The High Power Helicon (HPH) is a compact, electrode-less plasma propulsion device based on a radio frequency helicon discharge. Designing and constructing the most efficient thruster requires that the physics of the helicon wave be well understood. Currently it is unclear how energy from the antenna is coupled to the plasma to produce directed ion flow. HPH operates in at a different power level, pressure, frequency and magnetic field geometry than other helicon experiments. I will present measurements that demonstrate ion acceleration downstream of the helicon, which is not observed at other helicon experiments. I will also present measurements that demonstrate the influence of the helicon wave far downstream of the antenna. These measurements provide an explanation of the physics that drives the ion acceleration.
Bio: Jim Prager received his Bachelor of Science in physics from Lehigh University in Bethlehem, PA in January 2001. His senior research focused on methods of determining orbits of long-period binary star systems. In 2002 he moved to Seattle to begin graduate school in physics at the University of Washington. There he joined the Advanced Propulsion Laboratory to study radio frequency helicon waves in plasmas with applications to advanced space propulsion. He will receive his Ph.D. in June 2008.
Talk

Name: Aseem Agarwala, Adobe Systems, Inc.
Date: Wed, Apr 30 2008
Time: 4:00 - 5:00pm
Host: Jaeyeon Jung
Title: Matching the Mind's Eye: Getting more from our Photos and Videos
Abstract: Digital cameras are our primary tools for capturing the moments and memories of our lives. Our typical mental model is that photographs and videos are veridical records of what we saw with our own eyes. The reality, however, is that they are only interpretations of scenes, and they often fail to meet our expectations. In this talk I'll describe two very recent research projects that help us push our digital captures towards what we remember or hope to communicate. The first project looks at what large collections of photographs (e.g., Flickr) can tell us about the cameras that captured them. Specifically, we form novel statistical priors for large photo collections and use them to recover radiometric distortions of the appearance of a scene introduced by a specific camera model, such as a non-linear response curve, vignetting, and dead pixels on a camera sensor. These distortion parameters can then be used to undistort our photographs. In the second project, I describe a new way to perform local color and tonal adjustment within a single image or video. For example, a user may remember the scene as having a bluer sky, greener grass, and brighter people. The typical approach to making these adjustments within an image is to tediously mask out the separate regions. Instead, we allow the user to draw rough scribbles on different content and attach adjustments to them. Our algorithm then interpolates these adjustments to the rest of the image or video in a content-aware fashion. Specifically, we interpolate adjustments with an optimization that combines a boosting-based classifier on pixel appearance with an edge-weighted least-squares spatial regularization process
Bio: Aseem Agarwala is a senior research scientist at Adobe Systems, Inc. and an affiliate assistant professor at the University of Washington. He completed his Ph.D. in 2006 at the University of Washington, and his B.S. and M.Eng. at MIT in 1999. His work experience includes two years as a research scientist at Starlab in Brussels, Belgium, and research internships at Mitsubishi Electric Research Laboratory (MERL) and Microsoft Research. His areas of research are computer graphics, computer vision, and computational imaging. He received the Honorable Mention (runner-up) for the 2006 ACM Doctoral Dissertation Award, and a 2004 Microsoft Research Fellowship. His work can be also found in several products, including Adobe Photoshop and Adobe Photoshop Elements.
Talk

Name: Shwetak Patel, GaTech
Date: Tue, Apr 29, 2008
Time: 10:00 - 11:30am
Host: Anthony LaMarca
Title: Bringing Sensing to the Masses: An Exploration in Infrastructure Mediated Sensing
Abstract: The use of sensing systems in the home has the potential to impact various research areas such as chronic care management, aging in place, and sustainability. A major challenge remains in identifying and developing truly ubiquitous sensing solutions that address deployment challenges of cost-effectiveness, installation, maintenance, and overall acceptability for a general audience. In the home, the goal of practical ubiquity had led me to develop a new sensing approach, which I call "Infrastructure Mediated Sensing," or IMS. Infrastructure mediation refers to the simple augmentation and probing of existing home infrastructure, such as the electrical power lines, plumbing, or HVAC systems, to sense human activity. I will present three different IMS systems I have built that leverage the electrical and HVAC systems in a home for the purposes of location tracking and activity detection. I will describe an in-depth study of home mobility patterns enabled by an IMS-based positioning system, as well as motivate a wide variety of other applications this sensing approach enables. I will also describe research opportunities in exploring IMS outside of the domestic space.
Bio: Shwetak N. Patel is a Ph.D. candidate in Computer Science in the School of Interactive Computing at the Georgia Institute of Technology, where he is a member of the Ubiquitous Computing Research group, serves as the assistant director of the Aware Home Research Initiative, and is a National Science Foundation Graduate Research Fellow. His research is in the areas of Human-Computer Interaction and Ubiquitous Computing with a particular emphasis on developing and applying new low-cost, easy-to-use hardware and software solutions to enable novel application deployment and evaluation. Shwetak's published work has received various best paper awards and nominations. His past work on camera detection and neutralization received the designation of a Top Technology Idea of the Year from New York Times Magazine in 2005. Shwetak's research has also been the basis of various commercialization efforts.
Talk

Name: Helen Wang of Microsoft Research
Date: Wed, Apr 23, 2008
Time: 2:00 - 3:00pm
Host: Jaeyeon Jung
Title: Protection and Communication Abstractions for Web Browsers in MashupOS
Abstract: The advent of AJAX and client mashups has turned Web browsers into a multiprincipal operating environment. But browser support for Web programmers has lagged behind and remained in a single-principal world: The Same Origin Policy that dictates today's browser-security model offers either no trust through complete isolation between principals (sites) or full trust by incorporating third-party code as libraries. The consequences of such limited support include cross-site-scripting attacks that seriously plague today's Web and undesirable programming practices that make tradeoffs between security and functionality. In the MashupOS project, we address this deficiency. Our goal is to enable a browser to be a multiprincipal OS. Our initial focus is on protection and communication abstractions. Protection is to provide default isolation boundaries among principals (sites), while communication enables custom, fine-grained access control. We have designed our abstractions to be backward-compatible and easily adoptable. We have built a MashupOS prototype that we will demonstrate. Our experience and evaluation show that our abstractions make it easy to build more secure and robust client-side Web mashups and can be implemented easily in browsers with negligible performance overhead.
Bio: Helen J. Wang is a senior researcher and leads a security research group at Microsoft Research, Redmond. Her research interests are in system/network security, mobile/wireless computing, and wide-area large scale distributed system design. She received her Ph.D. degree from the Computer Science department of U. C. Berkeley in December, 2001. She obtained her Bachelor of Science in Computer Science from U. T. Austin, and Master of Science in Computer Science from U. C. Berkeley, respectively.
Talk

Name: Alexander Berg of Yahoo! Research
Date: Tue, Apr 22, 2008
Time: 10:00 - 11:30am
Host: Ali Rahimi
Title: Computational visual recognition
Abstract: Computational visual recognition concerns identifying what is in an image, video, or other visual data, enabling applications such as measuring location, pose, size, activity, and identity as well as indexing for search by semantic content. Recent progress in making economical sensors and improvements in network, storage, and computational power make visual recognition practical and relevant in almost all experimental sciences and many commercial applications including image search. My work in visual recognition brings together machine learning, insights from psychology and physiology, computer graphics, algorithms, and a great deal of computation.
I will present work on many aspects of attacking this challenge from low level image and video descriptors, to geometric models for deformable objects including humans, to techniques for parsing images of architectural scenes. This will include related work on modifying support vector machine approaches in order to increase recognition performance and speed on vision tasks. The applications in the presentation span object category recognition, image classification, action recognition, video search, biological monitoring, face recognition, and pedestrian detection.
Bio: Alex Berg's research concerns computational visual recognition. He is currently a research scientist at Yahoo! Research and a visiting scholar at U.C. Berkeley. He has worked on general object recognition in images, action recognition in video, human pose identification in images, image parsing, face recognition, image search, and machine learning for computer vision. His PhD at U.C. Berkeley developed a novel approach to deformable template matching. He earned a BA and MA in Mathematics from Johns Hopkins University and learned to race sailboats at SSA in Annapolis.
Talk

Name: Johnny Lee of CMU
Date: Tue, Apr 15, 2008
Time: 10:00 - 11:30 am
Host: Josh Smith
Tiitle: Enhancing the Practicality and Reachability of Interactive Technology
Abstract: As researchers, one of our common goals is to expand our reach and our capabilities as human beings through the development of new technologies and new ideas. Often we use immense resources to explore this new territory. However, an unfortunate side effect of this essential activity is that the number of individuals that can participate in this search becomes smaller and smaller as the resources become greater and greater. My primary motivation in research is to develop and demonstrate new techniques that substantially increase the practicality and reachability of technology. My work solves real world problems of applying research concepts by simplifying implementation and reducing system cost. This does two things: first, it enables more researchers to explore the domain advancing the state of research; second, it results in a more practical commercialization increasing distribution, adoption, and overall impact. In this talk, I will describe how I have successfully applied this philosophy in my research projects ranging from projector calibration, augmented reality, multi-touch interaction, immersive displays, animation, biometric interaction, to filmmaking.
Bio: Johnny Chung Lee is a PhD. Graduate student at Carnegie Mellon University in the Human-Computer Interaction Institute. His primary research interests are in developing technologies and techniques that increase the accessibility and practicality of technology. His previous work includes a diversity of topics ranging from projector calibration, augmented reality, brain-computer interfaces, haptic, animation, multi-channel audio, and filmmaking tools.
Talk

Name: Matt Welsh of Harvard, EECS
Date: Wed, Mar 26, 2008
Time: 4:00-5:00pm
Host: Jaeyeon Jung
Title: Fiji: A Platform for Data-Intensive Sensor Network Applications
Abstract: Sensor networks are becoming increasingly important for data-intensive applications that involve moderate to high data rates, fine-grained timestamping of recorded signals, and computationally-intensive processing within the network. Examples of this new class of applications include volcano monitoring, structural health monitoring, and biomedical data capture. In contrast to the first generation of sensor networks, which were focused on low-duty-cycle data collection and aggregation, these new applications demand much greater data fidelity and computational sophistication. At the same time, wireless sensor platforms are inherently resource-constrained, leading to severe limitations of computational horsepower, memory capacity, and radio bandwidth. The stringent application demands and resource constraints conflate to make programming complex sensor applications a very difficult task, even for experts in embedded systems. As a result there is a vast gap between the needs of domain scientists wishing to develop and deploy a sensor network and the level of expertise required to realize a resource-efficient implementation.
In this talk, I will present Fiji, a new programming platform intended to make it much easier for domain scientists to leverage wireless sensor networks. Fiji is based on the concept of macroprogramming, in which a program describing the global behavior of the network is compiled down to an efficient node-level binary. This is accomplished using a flexible dataflow-based intermediate form supported by multiple compilers for each target language. Fiji also provides a powerful node-level runtime and OS for resource-aware programming, allowing applications to naturally adapt to varying resource availability.
Bio: Matt Welsh is an associate professor of Computer Science at Harvard University. His research interests span many aspects of complex systems, including Internet services, distributed systems, and sensor networks. His current projects include macroprogramming language, operating system, and resource management techniques for sensor networks. He is the co-founder of AID Networks, an early-stage company developing wireless sensor platforms for emergency medicine. He is also a long-time Linux hacker and is the author of "Running Linux", published by O'Reilly and Associates.

Name: Bhaskara Marthi of MIT
Date: Tue, Mar 25
Time: 10:00 - 11:30am
Host: Matthai Philipose
Title: State estimation and decision making in complex systems
Abstract: I will describe an algorithm for probabilistic filtering, the problem of maintaining a probability distribution over the hidden state of a dynamical system given periodic noisy observations. This problem appears in various guises in practice, such as activity monitoring, state estimation, visual tracking, and fault diagnosis. Our algorithm, known as decayed MCMC, scales better than exact methods on many problems, and is less susceptible to losing track of the mode than the popular sequential Monte Carlo or particle filtering methods. Standard Markov chain Monte-Carlo mixing time analyses are insufficient to bound the complexity of our algorithm, and so we extend them to the setting of convergence of a marginal distribution. Time permitting, I will also briefly describe some other recent work on reinforcement learning with partial programs, and hierarchical planning for robotic manipulation.
Bio: Dr. Bhaskara Marthi is currently a postdoctoral research associate at MIT, working with Leslie Kaelbling and Tomas Lozano Perez on hierarchical planning and robotic manipulation. He received his PhD in 2006 from the University of California, Berkeley, working with Stuart Russell on reinforcement learning with partial programs, and its application to AI design for large real-time strategy video games. His other interests include probabilistic reasoning, relational and first-order models, and Monte Carlo algorithms.
Talk

Name: Tawanna Dillahunt of CMU
Date: Wed, Mar 12, 2008
Time: 4:00-5:00pm
Host: Beverly Harrison
Title: Leveraging internet scale technologies to help individuals to reduce energy consumption
Abstract: The average American consumes 12.5 times the energy of the average citizen of Africa or Asia. With approximately 300 million citizens, that adds up to 2 billion metric tons of CO2, or just over a 3rd of the total waste produced across all sectors of the U.S. economy. Of the other two thirds, much is produced as a by product of the process of meeting the needs of individuals. For example, the transportation of food, production of goods, and so on all generates waste. As a result personal choice can have a huge impact on energy consumption and waste production. The impact of individual choice can be seen in the increasing popularity of organic foods, hybrid cars, and other environmentally friendly consumption choices. What role can social technologies play in supporting large-scale group action and change?
Tawanna will discuss Footprints, a project aimed at encouraging sustainable behaviors through the use of social technologies. She will discuss StepGreen, a website that allows individuals to report and track their environmental impact and displays the results on popular social networking sites like Facebook and Myspace. She will then discuss the results of a pilot study of the motivational value of emotional attachment to a virtual polar bear. Finally, Tawanna will talk about UbiGreen, a collaboration with Intel, the University of Washington, and Carnegie Mellon to explore how concepts like the polar bear can be used in a mobile setting to motivate change. UbiGreen is a mobile application designed which displays feedback about environmentally sustainable transportation choices using either a polar bear or a tree. Tawanna will discuss the field study of UbiGreen that we are currently planning, and seek feedback on our future directions.
Bio: Tawanna is a first year Ph.D. student at Carnegie Mellon's Human Computer Interaction Institute and is advised by Jennifer Mankoff. Tawanna received her undergraduate degree from North Carolina State University in Computer Engineering. She completed her masters in Computer Science, with an an emphasis in Human Computer Interfaces from the Oregon Graduate Institute at the Oregon Health and Science University while working full-time as a Software Engineer at Intel. Tawanna worked at Intel for 7 years before starting at Carnegie Mellon. Her interests lie in using social computing to motivate and encourage positive behaviors and ubiquitous computing.
Talk

Name: Ara Knaian of MIT
Date: Tue, Mar 11, 2008
Time: 10:00 - 11:30am
Host: Josh Smith
Title: Electromagnetics and Acoustics for Ubiquitous Computing
Abstract: will present several projects on the theme of using electromagnetics and acoustics for sensing, actuation, communications, and power delivery for ubiquitous computing and robotics. I will describe my ongoing thesis work, on CMOS micro-robotics. We are constructing a millimeter-scale three-axis motion stage, with millimeter-scale travel, for applications in programmable matter and automation for biology and chemistry. The system uses custom-designed high-voltage CMOS IC's, which are designed to move when the dice are placed face-to-face. Each chip has a square array of micron-scale electrodes, which can be switched to +40V, grounded, or placed in a high-impedance input state. By driving and reading out the state from these electrodes, electrostatic actuation, communication, localization, and wireless energy transfer from IC to IC should be possible. E Ink Corporation developed and now manufactures electrophoretic displays for electronic books. Unlike liquid-crystal displays, electrophoretic displays are bistable and have a Lambertian reflectance characteristic, leading to low system power consumption, sunlight readability, wide viewing angle, and ink-on-paper aesthetic appearance. Because electrophoretic displays are bistable, areas of the display with differing switching history require differing voltage or pulse width to be switched to the same reflectance. If waveforms that account for differences in switching history are not used, artifacts such as "ghosting" result. I will describe my work on drive waveforms and
display controllers to account for switching history, and describe the algorithms and metrology equipment we developed to optimize these waveforms.
The MIT Autonomous Underwater Vehicle Team, Project ORCA, competes in an annual competition sponsored by the Office of Naval Research. As a founder and member of the team, I built a three-dimensional underwater acoustic direction finder, which enabled our vehicle to autonomously locate and center over a submerged acoustic beacon. The system used four hydrophones in a pyramidal array, coupled to a digital signal processor which performed pair-wise cross-correlation to recover time delays and geometric calculations to identify the direction of the incoming wave front. Finally, I will sketch some project ideas, including wireless power for mobile electronics and electromagnetic localization of home robots.
Bio: Ara Knaian is a PhD candidate in the Department of Electrical Engineering and Computer Science at the Massachusetts Institute of Technology. His is interested in the application of electromagnetics to problems in robotics and human-computer interaction. He also has interests in distributed computing, machine design, and computational geometry. Before returning to graduate school, Ara Knaian worked for E Ink Corporation, where he developed the electrophoretic display drive waveforms used in the SONY Reader and Amazon Kindle.
Talk

Name: Miryung Kim of UW
Date: Wed, Mar 5, 2008
Time: 4:00 - 5:00pm
Host: Jaeyeon Jung
Title: Analyzing and Inferring the Structure of Code Changes
Abstract: There is a significant gap between how programmers think about code changes and how change is represented in most change-centric software engineering tools such as diff, CVS, and Unix patch. To bridge this gap, I developed a new program differencing approach that automatically extracts a high-level change description from two program versions. The core of this approach is a novel rule-based change representation that explicitly and concisely captures systematic changes to a program's structure and a rule learning algorithm that automatically infers such rules. In this talk, I will also present my empirical studies on duplicated code, which partially motivated my program differencing approach. It has been long believed that code clones---syntactically similar code fragments---indicate bad smells of poor software design and that refactoring code clones improves software quality. By analyzing how code clones actually change over time, I found that code clones are not inherently bad and that immediate and aggressive refactoring may not be the best solution for managing code clones.
Bio: Miryung Kim is a Ph.D. candidate working with Dr. David Notkin at the University of Washington in Seattle. She earned her Bachelor's degree at the Korea Advanced Institute of Science Technology in 2001 and her Master's degree at the University of Washington in 2003. Her current research interests are software evolution, mining software repositories, and human aspects of software development.
Talk

Name: Yang Li of UW
Date: Wed, Feb 27, 2008
Time: 4:00 - 5:00pm
Host: Beverly Harrison
Title: Rapid Prototyping of Ubiquitous Computing Applications: Tools & Frameworks
Abstract: Pervasive or ubiquitous computing (ubicomp) applications can support people's everyday activities in the physical world by leveraging advances in sensor technologies and computing infrastructures. Designing ubicomp applications is challenging because our everyday activities are more complex, dynamic and less structured than the tasks supported by traditional desktop computing. Ubicomp design is difficult, time-consuming, and requires a high level of technical expertise, especially with sensor technologies. To address this, I created a set of rapid prototyping tools and frameworks. My early work with Topiary introduces high-level abstractions, such as maps and scenarios, for designers to easily model location contexts and specify location-based behaviors. Topiary also allows a design to be tested in the field via a Wizard of Oz approach, without deploying a location sensor infrastructure. My recent work is focused on activity-based ubicomp prototyping, a process for enabling long-term activities (such as keeping fit)-a larger unit for design than the tasks that are the focus of traditional design. To support such a process, I created ActivityDesigner, a system that allows designers to create functional prototypes of ubicomp applications based on field observations, and easily deploy and test these prototypes in situ.
Bio: Yang Li is a research associate in the Computer Science and Engineering Department at the University of Washington. He works in the areas of human-computer Interaction and ubiquitous computing, focusing on activity-based ubiquitous computing, rapid prototyping tools and pen-based interaction techniques. Previously, he was a postdoctoral researcher in EECS at the University of California at Berkeley. He received his PhD in computer science from the Chinese Academy of Sciences. http://www.cs.washington.edu/homes/yangli
Talk

Name: Richard Davis of University of California, Berkeley
Date: Wed, Feb 13, 2008
Time: 4:00-5:00pm
Host: David Wetherall
Title: K-Sketch: A "Kinetic" Sketch Pad for Novice Animators
Abstract: Because most animation tools are complex and time-consuming to learn and use, most animations today are created by experts. To help novices create a wide range of animations quickly, we have developed a general-purpose, informal, 2D animation sketching system called K-Sketch. Field studies investigating the needs of animators and would-be animators helped us collect a library of usage scenarios for our tool. A novel optimization technique enabled us to remove unnecessary complexity from the interface. The result is a pen-based system that relies on users' intuitive sense of space and time while still supporting a wide range of uses. In a laboratory experiment that compared K-Sketch to a more formal animation tool (PowerPoint), participants worked three times faster, needed half the learning time, and had significantly lower cognitive load with K-Sketch. Participants reported that they were no less comfortable showing their animations to others, and that they were more comfortable creating K-Sketch animations in front of others.
Bio: Richard Davis is a candidate for a Ph.D. in Computer Science from the University of California at Berkeley with a focus on Human-Computer Interaction. He specializes in systems that help everyday computer users express and manipulate rough ideas, including animations (K-Sketch) and notes taken in meetings (NotePals) or while listening to voice mail (Jotmail). His SketchWizard system helps designers simulate rough ideas for pen-based interfaces. In industry, he developed systems that helped people manipulate whiteboard notes (mimio), circuit signals (Simulink), microchip design information (Intel), and video productions (BorisFX). Richard earned bachelor's and Masters degrees in Electrical Engineering and Computer Science (with a minor in Theater Arts) from the Massachusetts Institute of Technology in 1995. He currently resides in Seattle and works with his research advisor James Landay at the University of Washington. More information on his current and past projects is available at
http://www.eecs.berkeley.edu/~rcdavis.
Talk

Name: Emily Cooper of Alium Labs
Date: Tue, Feb 5, 2008
Time: 10:00-11:30am
Host: Josh Smith
Title: Sensors, Miniaturization, and Sensor Miniaturization
Abstract: Miniaturized sensing systems can enable novel detection methods to drive research science and new applications through reduced package size and greater integration. However, shrinking feature sizes raises new challenges as devices become more sensitive to materials properties --- especially edge effects --- and geometric constraints imposed by fabrication methods. This talk will present approaches to sensing and miniaturization challenges through case studies from biosensing and navigation, including:
- fields and forces in biomolecular sensing: managing the charge microenvironment for improved potentiometric sensing
- applying optical interferometry to accelerometry
- a case against MEMS (MicroElectroMechanical Systems) implementation: miniaturizing a fluxgate magnetic field sensor
Bio: Emily Cooper received her PhD in Electrical Engineering at the Massachusetts Institute of Technology, where her thesis work focused on the development of novel detection strategies for bioscience. Working at the multi-disciplinary MIT Media Lab, she designed, fabricated, and applied MicroElectroMechanical (MEM) sensors, developed microfluidic systems for assay handling, and applied scanning probe techniques to nanolithography and surface analysis. After completing her graduate work, she joined TIAX, formerly the Technology and Innovation division of Arthur D. Little, where she developed applications in healthcare, portable power, data storage, environmental sensing, and navigation. Dr. Cooper currently runs a small technical and strategic consultancy.

Name: Kevin Fu of UMASS Amherst
Date: Thur, Jan 31, 2008
Time: 1:30-2:00pm
Host: Jaeyeon Jung
Title: Maximalist cryptography and computation on the WISP UHF RFID tag.
Abstract: With continuous improvements in the efficiency of microelectronics, it is now possible to power a general-purpose microcontroller wirelessly at a reasonable range. Our implementation of RC5-32/12/16 on the WISP UHF RFID tag shows that conventional cryptography is no longer beyond the reach of a general-purpose UHF tag. In this paper, (1) we provide preliminary experimental data on how much computation is available on a TI MSP430F1232 microcontroller-based RFID tag containing approximately 8~KBytes of flash and 256~bytes of RAM, and (2) we show that symmetric cryptography is feasible on an RF-powered, general-purpose RFID tag --- providing the first implementation of conventional cryptography on an RF-powered UHF RFID tag as far as we are aware.
Bio: Kevin Fu is an assistant professor in the Department of Computer Science at the University of Massachusetts Amherst, and is the principal investigator of the RFID Consortium on Security and Privacy (RFID CUSP). Kevin investigates the security and privacy of pervasive and invasive computation --- including RFID, implantable medical devices, and file systems. Kevin's contributions include key regression for efficient decentralized access control of storage; the SFS read-only file system for fast integrity-protected content distribution; proxy re-encryption file systems for managing distributed access control; and the security analysis of RFID-enabled credit cards, Web authentication, and software updates. Kevin received his M.Eng. and Ph.D. in Electrical Engineering and Computer Science at the Massachusetts Institute of Technology in 1999 and 2005 respectively, and his S.B. in Computer Science and Engineering from MIT in 1998. He has served on numerous program committees of prestigious conferences in computer security and cryptography. His research has appeared in The New York Times and The Wall Street Journal. Kevin also holds a certificate of achievement in artisanal bread making from the French Culinary Institute.
Talk

Name: Oren Etzioni of UW, CSE
Date: Wed, Jan 30, 2008
Time: 4:00-5:00pm
Host: Matthai Philipose
Title: Everything I know I Learned from Google: Machine Reading of Web Text
Abstract: Is it possible to capture a massive body of high-quality knowledge from the Web? my talk will describe our KnowItAll research project, which has been investigating this and related questions over the last five years. We have scaled and generalized information extraction methods to process arbitrary Web text, and to handle unanticipated concepts, but many challenges remain. One the most formidable challenges is moving from extracting isolated nuggets of information to capturing a coherent body of knowledge that can support automatic inference. My talk will highlight this and other exciting directions for future work.
Bio: Oren Etzioni received his bachelor's degree in Computer Science from Harvard University in June 1986 where he was the first Harvard student to "major" in Computer Science. Etzioni received his Ph.D. from Carnegie Mellon University in January 1991, and joined the University of Washington's faculty in February 1991, where he is now a Professor of Computer Science. Etzioni received a National Young Investigator Award in 1993, and was selected as a AAAI Fellow a decade later. In 2007, he received the Robert S. Engelmore Memorial Award. He is the founder and director of the University of Washington's Turing Center.
His current research interests include: fundamental problems in the study of intelligence, Web search, Machine Reading, and data mining. Etzioni has been serving as a director of the non-profit AI Access Foundation since 1993. The foundation was created by Steve Minton to publish the Journal of AI Research --- one of the very first electronic journals distributed over the Web. Etzioni is an Associate Editor of the ACM Transactions on the Web.

Name: Nati Srebro of Toyota Technological Institute
Date: Wed, Dec 19, 2007
Time: 4:00 - 5:00 pm
Host: Tanzeem Choudhury
Title: Does a large data-set mean more, or less, work?
Abstract: In this talk we will consider how the computational cost of several machine learning tasks depends on the amount of available information (data set size). In devising methods for optimization problems associated with learning tasks, and in studying the runtime of these methods, we usually think of the runtime as increasing with the data set size. However, from a learning performance perspective, having more data available should not mean we need to spend more time optimizing. At the extreme, we can always ignore some of the data if it makes optimization difficult. But perhaps having more data available can actually allow us to spend less time optimizing? I will describe two types of behaviors:
(1) a phase transition behavior, where a computationally intractable problems becomes tractable, at the cost of excess information. I will demonstrate this through a detailed study of informational and computational limits in clustering.
(2) the scaling of the computational cost of training, e.g. support vector machines (SVMs). I will argue that the computational cost should scale down with data set size, and up with the "hardness" of the decision problem. In particular, I will describe a simple training procedure, achieving state-of-the-art performance on large data sets, whose runtime does not increase with data set size.
Joint work with Sam Roweis (U Toronto/Google), Gregory Shakhnarovich (Brown), Shai Shalev-Schwartz (TTI) and Yoram Singer (Google).
Bio: Nathan Srebro received his PhD from MIT in 2004. After visiting the University of Toronto and IBM Haifa Research Labs, he is now an assistant professor at the Toyota Technological Institute--Chicago (a philanthropically endowed academic computer science institute) and at the University of Chicago.

Name: Daniel Gatica-Perez of IDIAP Research Institute
Date: Tue, Dec 11, 2007
Time: 10:30-11:30am
Host: Tanzeem Choudhury
Title: Modeling social interaction in small group meetings
Abstract: In this talk, I will present ongoing work on automatic analysis of social interaction patterns in small group meetings from sensor data, in the context of a meeting room equipped with multiple cameras, microphones, and text capture devices. I will discuss statistical models to recognize (1) visual attention and (2) perceived dominance of meeting participants, which integrate specific aspects of the multiparty, multimodal nature of group conversations. I will discuss our experience on what perceptual cues - some of which have been borrowed from social psychology - have worked well (or not) for our goals, and on issues related to performance evaluation of methods for social interaction recognition and discovery.
Bio: Daniel Gatica Perez is a senior researcher at IDIAP Research Institute in Martigny, Switzerland, and a collaborator/lecturer at the Swiss Federal Institute of Technology in Lausanne (EPFL). His research interests include multimedia signal processing and information retrieval, social computing, and machine learning applied to these domains. He has worked on automatic modeling of human activity in the context of Swiss, European, and US funded research projects since 2002. He got a PhD in Electrical Engineering from the University of Washington in 2001, and currently is an Associate Editor of the IEEE Transactions in Multimedia.

Name: Dan Olsen of BYU
Date: Thur, Nov 15, 2007
Time: 3:00-4:00pm
Host: Beverly Harrison
Title: Cubic Inch Computing
Abstract: Advances in digital electronics have made computing smaller and cheaper. If your personal computer is less than 1 cubic inch in volume, how will you interact with it? The problem is that personal computing is rapidly dropping below the physical limitations of human beings. This talk will discuss the challenges of highly nomadic personal computing and the UI architectures that can overcome those limitations. Example implementations of nomadic interactions where interactive resources such as displays and input devices are annexed rather than carried will be presented.
Bio: Dan Olsen is a Professor and past Chair of Computer Science at Brigham Young University and currently directs the ICE (interactive computing everywhere) project there. He is a recognized expert in HCI, novel UI and interaction techniques, and applying machine learning for non-expert users. Dan founded and directed the HCI Institute at CMU, is an ACM Fellow, recipient of a CHI Lifetime Achievement award, Father of UIST award to name but a few. He is an active and visible member of most HCI conferences and committees, has published extensively, and was most recently invited to speak at the UW Distinguished Lecture series
Talk

Name: Maya Gupta of UW
Date: Wed, Nov 14, 2007
Time: 4:00-5:00pm
Host: Ali Rahami
Title: Recent Advances in Nearest Neighbor Learning: Weights, Neighbors, and Estimates
Abstract: Recent advances in nearest-neighbor learning are shown for finding neighborhoods, neighborhood weighting, and estimating given nearest-neighbors. In particular, it is shown that weights that solve linear interpolation equations minimize the first-order learning error, and when coupled with the principle of maximum entropy this results in significantly improved nearest-neighbor classification performance. We show how these weights are related to weights formed by local linear regression. New approaches to adaptively determining neighborhoods are discussed. Standard weighted nearest-neighbor estimation maximizes likelihood, and it is shown that minimizing expected Bregman divergence instead leads to optimal solutions in terms of expected misclassification cost.
Bio: Maya Gupta completed her Ph.D. in Electrical Engineering in 2003 at Stanford University as a National Science Foundation Graduate Fellow, after taking a BS in Electrical Engineering and a BA in Economics from Rice University in 1997. From 1999-2003 she worked for Ricoh's California Research Center as a color image processing research engineer. In the fall of 2003, she joined the EE faculty of the University of Washington as an Assistant Professor. She was awarded the 2007 Office of Naval Research Young Investigator Award, and the 2007 Univ. of Washington Outstanding Teaching Award.
Talk

Name: Meeyong Cha of KAIST
Date: Wed, Oct 17, 2007
Time: 4:00 - 5:00pm
Host: Jaeyeon Jung
Title: I Tube, You Tube, Everybody Tubes: Analyzing the World's Largest User Generated Content Video System
Abstract: User Generated Content (UGC) is re-shaping the way people watch video and TV, with millions of video producers and consumers. In particular, UGC sites are creating new viewing patterns and social interactions, empowering users to be more creative, and developing new business opportunities. In this talk, I will present the intrinsic statistical properties of UGC video popularity based on real traces from YouTube, the world's largest UGC sharing web site. Understanding the popularity characteristics is important because it can bring forward the latent demand created by bottlenecks in the system ( e.g., poor search and recommendation engines, lack of metadata). I will also discuss the potential for more efficient UGC VoD systems (e.g., P2P distribution, caching).
Bio: Meeyoung Cha is a PhD student in Computer Science at KAIST, Korea. Her advisor is Dr. Sue Moon. She is working on the network design and support for multimedia streaming services. Previously, she was an intern at AT&T Labs Research in NJ, where she participated in the cost comparison of IPTV backbone designs. Recently, she was an intern at Telefonica Research in Barcelona, Spain, and in University of Cambridge, UK, where she analyzed a nationwide IPTV system and the world's largest VoD for user-generated contents, YouTube. She also maintains interests in path diversity issues in intra- and inter-domain routing. She expects to graduate in Feb 2008.

Name: Phil Levis of Stanford University
Date: Wed, Sep 26, 2007
Time: 4:00 - 5:00pm
Host: Ben Greenstein
Title: Visibility: A New Metric for Protocol Design
Abstract: After nearly ten years of research, industrial development, and successful deployments, deploying wireless sensor networks remains difficult and labor-intensive. This encountered complexity is more than an artifact of dealing with a novel technology: energy constraints make it an essential consideration that will not go away. In practice, distributed algorithms, limited state, energy, and low-power local communication make it difficult to observe or understand a network's internal operations and decisions. This challenge has led to a variety of management and debugging systems, such as SNMS, Sympathy, Marionette, and Clairvoyant, all of which seek to give an administrator the ability to peek into the state of the network. In this talk, we propose a different approach. Instead of adding visibility layers on top of an inherently obfuscated system, we ask the question: "How can we design a network architecture to improve the visibility of its internal decisions?" We present the MNet architecture, a network architecture for wireless networks which has visibility as its major design principle. We define a quantitative measure for visibility. We describe the responsibilities of the MNet protocol stack, including the Fair Waiting Protocol (FWP), the architecture's narrow waist protocol that sits between multihop layers and a CSMA/CA link layer. We present the Pull Collection Protocol as an example of visibility-driven protocol design; to reduce failures, PCP shifts the communication burden from data sources to the data sink, which pulls packets in from the rest of the network. We conclude with comments on areas of current and future work in the MNet architecture.
Bio: Philip Levis is an Assistant Professor in the Computer Science and Electrical Engineering Departments of Stanford University. He researches embedded wireless networks, including programming languages, operating systems, network protocols, algorithms and applications. His prior work includes TOSSIM, the TinyOS simulator, the Trickle algorithm for data dissemination in wireless networks, application-specific virtual machines, sensornet OS power management, wireless measurement, and wireless protocol design. He is the chair of the Core Working Group of the TinyOS Alliance.

Name: Shaz Qadeer of Microsoft Research
Date: Wed, Aug 29, 2007
Time: 4:00 - 5:00pm
Host: Tanzeem Choudhury
Title: Iterative context-bounding: a new approach for finding errors in large multithreaded programs Abstract: Multithreaded programs are difficult to get right. The interaction of concurrently executing threads leads to a huge number of program behaviors. Programmers, unable to account for all possible interactions among threads, often make errors which are difficult to find by traditional testing methods. In this lecture, I will present CHESS, a software model checker for systematically enumerating such behaviors. CHESS implements iterative context-bounding, a new approach for effectively searching the state space of a multithreaded program. In an execution of a multithreaded program, a context switch occurs when a thread temporarily stops execution and a different thread starts. Iterative context-bounding gives priority to executions with fewer context switches, exploring all executions with no context switches followed by all executions with one context switch and so on. For a fixed number of context switches, the total number of executions in a program is polynomial in the number of steps taken by each thread. This theoretical upper bound makes it practically feasible to scale systematic exploration to large programs without sacrificing the ability to go deep in the state space. Our experience applying CHESS to large real-world programs shows that systematic search with a small number of context switches has the ability to expose nontrivial concurrency bugs. CHESS has uncovered 9 previously unknown bugs in our benchmarks, each exposed by an execution with at most 2 context switches.
Bio: Shaz Qadeer is a researcher in the Software Reliability Research group at Microsoft Research. The goal of his research is to develop tools for improving the productivity of programmers. He has worked on many program verification tools, spanning the range from run-time verification to model checking to static analysis, with a special emphasis on tools for concurrent programs. Shaz received his Ph.D. at UC Berkeley and worked at Compaq Systems Research Center before joining Microsoft Research.

Name: Ari Juels of RSA Laboratories
Date: Wed, Aug 22, 2007
Time: 4:00 - 5:00pm
Host: Tanzeem Choudhury
Title: RFID Security: In the Shoulder and on the Loading Dock
Abstract: RFID (Radio-Frequency IDentification) tags are microchips that communicate via radio. In common use today, they promise to become a ubiquitous tool for labeling objects and identifying people. RFID thus carries a strong imperative for protection against counterfeiting and privacy infringement. In this talk, I'll give a brief introduction to RFID use today and describe some challenges and potential solutions in special operating environments. I'll discuss human-implantable RFID devices and the intricate privacy and security problems associated with these "prosthetic biometrics." I'll also talk about the problem of secure key distribution, a perennial challenge in computing systems that is particularly thorny in RFID-enhanced supply chains.
Bio: Dr. Ari Juels is Chief Scientist and Director of RSA Laboratories, where he works to bring sparks of invention and insight from RSA's scientists and affiliates to the company as a whole. He joined RSA (now a division of EMC) in 1996 after receiving his Ph.D. in Computer Science from UC Berkeley.

Name: Jesse Walker of Intel Corporation
Date: Wed, Jul 25, 2007
Time: 4:00 - 5:00pm
Host: Ben Greenstein
Title: 802.16e security
Abstract: This talk discusses WiMAX security as defined by 802.16e. It examines the 802.16e security architecture, the data protection mechanisms, key management, and authentication procedures...
Bio: Dr. Jesse Walker is Intel's Network Security Architect. He is responsible for developing and proliferating Intel's guidelines to ensure secure networked communications with Intel-based devices. Dr. Walker is recognized as a security expert. He is the Technical Editor for 802.11 security enhancements, worked on the original 802.11 specification, and was the first person to publicly identify security vulnerabilities in the 802.11 WLAN protocol. Jesse has been with Intel for five years. Before joining Intel he worked at Shiva, Raptor Systems, Rockwell International, Datapoint Corporation, Iowa State University, and the University of Texas. Dr. Walker holds a Ph.D. in Mathematics from the University of Texas at Austin, and a B.A. in Liberal Arts (also from UT). Jesse has published extensively in academic and technical journals, including the May 2003 issue of Communications of the ACM. He has also performed a number of high-visibility speaking engagements.
http://www.intel.com/technology/techresearch/people/bios/walker_j.htm
Talk

Name: Wanda Pratt of University of Washington
Date: Wed, Jul 18 , 2007
Time: 4:00 - 5:00pm
Host: Sunny Consolvo
Title: Managing Health Information in Your Life
Abstract: As clinicians are forced to decrease time spent with patients and as the specialization and fragmentation of care increases, patients are required to play an increasingly active role in their health care. Yet, few information tools exist to support patients in this active role. Patients often must coordinate their health care across multiple clinicians, learn new health terminology, make treatment choices, manage their home care, track insurance benefits, etc. These information rich tasks demand information management work of patients. The long-term objectives of this research are both to understand patients' information management work and to develop new technology that will support that work. Thus, we hope to help patients actively participate in their health care as they maintain the personal and professional aspects of their lives. In this talk, I will connect some of my older research on search interfaces to this emerging work on personal health information management.
Bio: Dr. Pratt is an Associate Professor in both the Information School and the Division of Biomedical & Health Informatics in the Medical School at the University of Washington. She is also the Director of the Graduate Program in Biomedical and Health Informatics. She received her Ph.D. in Medical Informatics from Stanford University, her M.S. in Computer Science from the University of Texas, and her B.S. in Electrical and Computer Engineering from the University of Kansas. Her published papers span a wide range of topics whose central theme is to understand the problem of information overload in a variety of health contexts and to develop new types technology to address those problems. She received an NSF CAREER Award for her work on literature-based discovery systems, is on the editorial board for the Journal of Biomedical Informatics, and serves on the standing NIH grant-review committee for the National Library of Medicine.
Talk

Name: Mitch Lum of University of Washington
Date: Wed, Jul 18 , 2007
Time: 4:00 - 5:00pm
Host: Josh Smith
Title: NASA NEEMO XII participation
Abstract: Telemedicine is already changing the means by which patients receive healthcare. Telesurgery will provide new opportunities for surgical intervention. Mobile Robotic Telesurgery (MRT) particularly in Extreme Environments has the potential to deliver emergency medical care to the critically injured; be it soldiers on a battlefield, civilians in a disaster zone or even astronauts in a deep space mission. The University of Washington, BioRobotics Lab has developed a new MRT platform, the RAVEN. In collaboration with the University of Cincinnati, the RAVEN was tested in the National Oceanic and Atmospheric Administrations (NOAA) Aquarius Undersea Habitat as part of the NASA Extreme Environments Mission Objective 12 (NEEMO 12). Surgeons located in Seattle, WA teleoperated the RAVEN, located at a depth of 60 feet inside the Aquarius Habitat, three and a half miles offshore from Key Largo, FL. This experiment tested both the RAVENs ability to operate in an extreme environment as well as the surgeons' skill in operating under variable time delay. Further this mission demonstrated the ability to use a single platform for multiple objectives, including educational outreach and geological sample handling and analysis.
Bio: Mitchell Lum is a PhD candidate in the University of Washington, Dept. of Electrical Engineering, BioRobotics Lab. He joined the lab as a sophomore undergraduate in 1999. In 2002 he finished his BSEE and continued with the UW BioRobotics Lab for his graduate work, on a research grant from the US Army Medical Research and Materiel Command to create a new surgical robot system. As a research assistant on the RAVEN Surgical Robot he has been involved with its development from concept to its current state, having led the team on two successful mobile robotic telesurgery (MRT) missions.
Talk

Name: Huong Q. Nguyen of University of Washington
Date: Wed, Jun 27, 2007
Time: 4:00 - 5:00pm
Host: Sunny Consolvo
Title: Utility of Technology-Enabled Interventions to Support Sustained Exercise
Abstract: Despite optimal medical therapy, patients with chronic obstructive pulmonary disease (COPD) continue to experience dyspnea (shortness of breath) with their activities of daily living. Exercise is one such evidence-based intervention that is effective in reducing dyspnea and improving functional capacity. However, exercise persistence for patients with COPD can be especially challenging given the chronic progressive nature of the illness. The thoughtful integration of information and communication technologies for tailored, real-time support may help patients persist with exercise over time. This presentation will cover the following areas: 1) development and current implementation of a PDA-based exercise intervention for patients with COPD; 2) lessons learned and opportunities for improvement and expansion to a broader older adult population and 3) potential collaborative research projects and funding mechanisms.
Bio: Dr. Nguyen is currently an Assistant Professor in the Department of Biobehavioral Nursing and Health Systems at the University of Washington. Her research program has been focused on developing and testing technology-enabled interventions to support self-management in older patients with chronic illnesses. She is currently involved in two NIH funded projects that are evaluating interventions to support exercise persistence in patients with chronic lung disease. Dr. Nguyen is also receiving training as a K12 Multidisciplinary Clinical Research Scholar.
Talk