Tai-Sing Lee Associate Professor, Computer Science and the CNBC Carnegie Mellon University Phone: (412) 268-1060 Fax: (412) 268-5060 Email: tai@cnbc.cmu.edu Individual Website: http://www.cnbc.cmu.edu/~tai Lab Website: http://www.cnbc.cmu.edu/apl/

Ph.D., Harvard University Research Interests My research involves the application of computational, modeling and electrophysiological techniques to study the neural basis of visual perception and recognition. The current effort of my laboratory is focused on the following scientific issues in the areas of computational neuroscience and computational vision. The first issue concerns feedback and hierarchical computation in the visual system. The classical paradigm for vision, as delineated by Marr, is that vision is accomplished by a series of feedforward computations in the visual hierarchy. The experimental findings from my laboratory show that global contextual information can modify the computation in early visual areas, presumably mediated by the massive recurrent feedback from the higher level cortices to the earlier ones. What is the functional role of this feedback? What are the advantages of the concurrent and interactive computation across the visual hierarchy? To address these questions, we are constructing and testing computational models and realistic neural circuits in conjunction with recording from neurons across the visual hierarchy while the monkeys are observing stimulus patterns of different complexity. The second issue concerns the dynamic and active aspects of vision. Much of the vision research has focused on the analysis of patterns in static situations by the visual system or by computer. Vision in reality is an active and dynamic process. Our eyes move constantly, purposefully scanning the environment to construct a coherent internal representation of the scene based on the retinal data, which is precise only in the fovea. How are these bits and pieces of retinal information assembled and integrated over time to form a seemingly coherent picture of the world? How and where are these scene elements represented in the brain? To address these questions, we are conducting human psychophysical experiments to examine information transfer across saccadic eye movements, formulating and testing computational models for scene integration, recording from neurons when monkeys are actively scanning and searching in the visual environment. The third issue concerns neural plasticity and learning. The brain is an adaptive system. Even after development, the neural circuits remain plastic and exhibit changes with learning. Currently, we are implanting electrode arrays of 100 electrodes into the monkeys' neocortex to record from the same neurons in the visual system over long periods of time (months) during which the monkeys will be trained to perform new perceptual grouping and perceptual discrimination tasks. The experiments are inspired by unsupervised and reinforcement learning models, but we hope to infer from the neurophysiological data the neural algorithms underlying the development, formation and maintenance of neural circuits in the visual hierarchy. The fourth issue concerns the structure of neural code. For the last thirty years, the average firing rate of the neurons have been considered to be the most reliable measure of neural information. We are analyzing single and multi-electrode data to examine the possible existence of precise temporal spike patterns hidden in the neural spike trains that may encode higher order structures using various pattern analysis, machine learning and statistical techniques. By elucidating the neural code, representations and algorithms underlying biological computation and learning, we hope not only to gain a better understanding of the mind and the brain, but also to discover new and more powerful ways to build learning systems and robots. Recent Publications Lee, T.S., Yang, C., Romero, R. and Mumford, D. (2002) Neural activity in early visual cortex reflects experience and higher order perceptual saliency. Nature Neuroscience, 5(6), 589-597. Lee, T.S. & Nguyen, M. (2001). Dynamics of subjective contour formation in early visual cortex. Proceedings of the National Academy of Sciences, U.S.A., 98(4) 1907-1911}. Lee, T.S., D. Mumford, R. Romero and V.A.F. Lamme (1998). The role of primary visual cortex in higher level vision. Vision Research 38, 2429-2454. Lee, T.S. (1996). Image representation using 2D Gabor wavelets. IEEE Transection of Pattern Analysis and Machine Intelligence. Vol. 18, No. 10, October, 959-971. Lee, T.S. (1995). A Bayesian framework for understanding texture segmentation in the primary visual cortex. Vision Research 35, 2643-2657..