Chris Halter and Jim Levin, Teacher Education Program, UCSD

DIVE-In to Teaching: Using Digital Video Tools to Support Teacher Development
A look at scaffolding pre-service teacher development through critical self-reflection and peer collaboration through the use the Internet and DIVER, a digital video analysis tool.

Morana Alac, Dept of Cognitive Science, UCSD

Scientists using fMRI technology strive to design and carry out experiments in which subjects' movements are reduced to minimum. Such movements produce artifacts that are seen as obstacles for successful research. I examine how the imagined causes of these artifacts are re-produced during data analysis as part of discursive and embodied social action and what their role is in knowledge production. In order to trace the ways in which these elements leak back into scientific practice, I focus on traditionally less explored communication modalities, such as non-linguistic sounds, body movements, gestures, etc.. I describe how, in two instances of fMRI apprenticeship, such semiotic modalities are combined with brain images to produce understanding. The video-taped interaction reveals how phenomena of interest are conceptualized as dynamic and hybrid events generated through schematic coordination of different semiotic modalities.

During the second part of the meeting Hollie Crower and I will show you David Byrne's DVD Envisioning Emotional Epistemological Information on his use and abuse of Power Point.

Tim Marks, Dept of Cognitive Science, UCSD

G-flow: A generative model for fast tracking using 3D deformable models
Tracking an object (such as a human face) in 3D from 2D video data is a difficult problem that is crucial to many potential applications of computer vision. The problem is difficult because the observed variation in the 2D images over time can be caused by a variety of sources in the (three-dimensional) world, such as rigid motion, non-rigid motion, and changes in lighting. Existing approaches to object tracking in the computer vision literature can be divided into motion-based (or optic flow-based) approaches and template-based approaches. Flow-based methods implicitly assume good knowledge about the location of the object at each time step, and thus have a tendency to drift as errors accumulate. Initialization and recovery from drift are open issues in motion-based approaches, and they are typically handled using heuristic methods. At the other end of the spectrum, template-based approaches assume good knowledge about the appearance of the object of interest. Template-based approaches have difficulty dealing with realistic sources of variation (pose, illumination, identity, expression, etc), and typically rely on heuristics that allow for dynamic updating of the templates and periodic re-registration. In practice, the issues of model initialization, dynamic updating of templates, error detection, and re-initialization are still unsolved. Finding principled solutions to these problems is critical to the widespread application of this computer vision technology in daily life. I will present a generative model (G-flow) and inference algorithm for simultaneous tracking of 3D pose (rigid motion), non-rigid motion, object texture, and background texture. Under this model, optimal inference about pose and texture can be performed efficiently using a bank of Kalman filters for texture whose parameters are updated by an optic-flow-like algorithm. The inference algorithm unifies optic flow-based and template-based tracking methods, dynamically adjusting the relative importance of each component in a principled manner, and elucidating the conditions under which each type of algorithm (flow-based or template-based) is optimal. I will demonstrate a non-rigid face tracking system based on G-flow.

Jean-Baptiste Haué, Post-doc in D-COG/HCI and Interactive Cognition Labs, UCSD

Bringing meaning in driver’s study
The Nissan Project gathers seven universities working with the NRC (Nissan Research Center) in Japan. It is in its 3rd year and just had its 4th workshop. The team from the HCI-D-Cog lab participates to this project by 1) Building a database of natural drive situations to be shared by all the participants of the project, 2) Bringing an ethnographic approach to analysis and 3) Providing some recommendations for Design. Several issues are brought by study. One is on what ethnography can bring to the others approaches and to the designers. It appears that a description respecting the meaning of the driver during natural situation is complementary to the theoretically driven approaches and to the simulator studies. The ethnographic analyses of the common notions, like lane changes, show the importance of the context from the road and the traffic but also from the driver’s preferences and feelings. The NRC is also concerned about the rejection of some systems by the consumers and wants to have some feed back from real users. A tool is necessary to browse the huge amount of data collected before, during, and after the drive. Erwin Boer, coordinating the universities and the NRC, provide us Matlab tools to have an environment for ethnographic analysis combined with automated processing of data. Some algorithms for automatic detection and visualization will also be provided from others labs. These tools address the more general issues of information visualization and pattern research. Eventually, the need to formalize the ethnographic knowledge leads to build taxonomy of situations and to identify patterns of behavior. The use of categories shared with cognitive tasks decomposition and studies of time of behavioral responses allow a discussion with theses disciplines. But at any time it is possible to come back to a detail analysis to see what is going on for the driver. In Wednesday talk, after a personal presentation, the Nissan project and the work from the HCI-D Cog lab will be presented.

Bio
Jean-Baptiste Haué is post-doc at UCSD from June 2003. He is sharing his time between the ICLab and the HCI-D-Cog lab. At the ICLab he is working on collaborative work in small team, in face to face / distributed environment. At the HCI/D-Cog lab, he is working on the ethnography of drive activity in the Nissan project. After his engineering diploma at the UTC (Université de Technology de Compiègne), in France, he has done his PhD at the R&D department of EDF (Electricité de France) supervised by Jacques Theureau from the UTC. In the Human Computer Interaction group he has worked on Cognitive Engineering methodology applied to heating systems, about the construction of HCI specifications from ethnographic knowledge.

François Guimbretière, Human-Computer Interaction Lab (HCIL), Univ. of Maryland

People, Paper and Computers
For several decades, experts have predicted that the advent of more powerful and compact computers will result in the creation of paperless offices. Yet, the consumption of paper is on the rise and, with few exceptions, office work still relies heavily on paper. At the root of this apparent paradox is the tension between the set of affordances provided by printed and digital documents: on the one hand, printed documents are easy to navigate, annotate and provide large inexpensive high-resolution display surfaces. Their tangibility also makes them easy to navigate. On the other hand, digital documents are easy to edit, search and index. Their intangibility makes them inexpensive to store, duplicate and distribute. The project on People, Paper and Computers explores how to design new human computer interfaces that will bridge the affordance gap between printed and digital documents. These interfaces will let users navigate and annotate digital documents with the ease and comfort of printed documents. In this talk, we will present an overview of the project on People, Paper and Computers and report on the current status of several major components including the Paper Augmented Digital Document system.

Bio
François Guimbretière is an assistant professor at the University of Maryland Human-Computer Interaction Lab (HCIL). His current research interests include exploring how new technologies can be used to reduce the gap between the digital world and the paper world; designing and quantifying new command selection mechanisms such as FlowMenu and understanding how new interaction and rendering techniques could help people understand and compare very large trees such as phylogenies. More information can be found at http://www.cs.umd.edu/~francois.

Amaya Becvar, D-COG/HCI Lab, UCSD

Digital Pen/Analog Pen - Enhancing the cognitive elegance of paper-based media
For several decades, visionaries have predicted a move away from an analog paper-based workplace to a digital workplace. However, even with the advent of fast, cheap computers and the World Wide Web, paper consumption has instead exponentially exploded. It seems that people prefer paper for many tasks, because paper has key affordances as yet unrealized in digital media. But digital media has important advantages that paper does not, such as the ability to be rapidly searched and indexed. To this end, designers have worked for many years trying to build digital devices that mimic the affordances of paper while maintaining the advantages of digital technology. However, these devices have met with little success in real-world workplaces. Paper is just too ubiquitous, inexpensive, and easy to use. Perhaps the answer is not to build digital tools that copy the affordances of paper, attempting to "inject" them into a functional workplace, but rather to work on building tools that can bring paper and computer together, so that people can effortlessly mix between the two media. The digital Anoto pen is a device attempting to provide pathways between the analog and the digital world. The pen looks and feels like a real ballpoint pen, but has a system to automatically record pen strokes on paper. In this talk, I will introduce the Anoto pen technology, situated in a discussion of cognitive design. I will then discuss some potential applications of the Anoto pen in educational and workplace settings.

William J. Clancey, Chief Scientist, Human-Centered Computing Computational Sciences Division, NASA-Ames Research Center (Moffett Field, CA)

Saeko Nomura, Visiting Post-doc in D-COG/HCI Lab, UCSD

Natalie Jeremijenko, Dept. of Visual Arts, UCSD

On designing tangible media and structuring participation
Can we really design Interaction? How do we reconcile an agent performing particular tasks for which a system is designed, with the informal, iterative and cumulative actions that constitute interactions in situated contexts? I present an theoretical device, empirical study and example designs that explore an approach that has descriptive plausibility for both contexts. In particular I argue that changing the unit of analysis from individuals interacting with devices, to the structure of participation of small group contexts, allows us to evaluate designs more effectively with respect to particular social institutions. This will allow us to design and evaluate systems in which we are concerned less with the series of interactions between an agent and a device, and more with the capacity of the agent to understand and act in the techno-social context that the device provides.

Bio
Natalie Jeremijenko, is a design engineer and technoartist who has recently joined the faculty in Visual Art, UCSD. Prior to this she has been on the Faculty of Engineering, Dept of Mechanical Engineering, Yale University. She has been named one of the top one hundred young innovators by the MIT Technology Review, and her work has been featured in the Tate Gallery Cream 2, and a large project was commissioned for the opening of the museum MASSMoCA (www.massmoca.org ). Jeremijenko's projects include digital, electromechanical, and interactive systems in addition to biotechnological work. These have also been screened and exhibited at the Guggenheim Museum, the Museum Moderne Kunst, Frankfurt, the LUX Gallery, London, the Whitney Biennial, Documenta, Ars Electronic prix, presented at the Museum of Modern Art in New York. She was a 1999 Rockefeller fellow. She did graduate engineering studies at Stanford University in Mechanical Engineering, and at the University of Melbourne in the History and Philosophy of Science Department and her Ph.D. is in the Dept of Information Technology and Electrical Engineering, University of Queensland. As the director of the Engineering Design Studio at Yale University she developed and implemented new socio technical courses in technological based innovation. She is also affiliated with the Media Research Lab/Center for Advanced Technology in the Computer Science Dept., NYU, where she did postdoctoral studies. Other research positions include several years at Xerox PARC in the computer science lab, and the Advanced Computer Graphics Lab, RMIT. She has also been adjunct faculty in digital media and computer art at the School Of Visual Art, New York and the San Francisco Art Institute. She is known to work for the Bureau of Inverse Technology.

Research Interest
Tangible Media, Physical Computing, Human Computer Interaction, Distributed Sensing, Robotics, interaction design for nonhumans, Ubiquitous Computing, Innovation Processes, Interdisciplinary Knowledge transfer, Lay/Expert Knowledge, Engineering and Technical Education, Characterizing technosocial change, Toy Design, Sociology of Scientific Knowledge, Material Culture, Smart Building Design