Feb. 17, 11:45, Wei Ji Ma. Optimality and Probabilistic Computation in Visual Categorization

Participating Faculty


 Kostas Bekris

Assistant Professor


Computer Science


I am interested in designing algorithms for planning the motion of physically-grounded agents, including robots which effectively interact with humans, as well as virtual agents in physically realistic and interactive simulations. The objective is to increase the capabilities, robustness and safety of physical systems, or the physical realism of simulated agents, while aiming for real-time performance. Perceptual science lies at the core of dealing with the interaction of robots and people as well as the design of effective interacting simulations.

Kostas Bekris received his BS degree (2001) in Computer Science at the University of Crete, Greece and completed his MS (2004) and PhD degrees (2008) in Computer Science at Rice University, Houston, TX, under the supervision of Prof. Lydia Kavraki.  He was an Assistant Professor and co-director of the Robotics Lab at the University of Nevada, Reno between 2008 and 2012. His research group has been supported by the National Science Foundation, the National Aeronautics and Space Administration, the Office of Naval Research and the Nevada NASA Space Grant Consortium. Kostas is co-chair of the IEEE Robotics and Automation Society Technical Committee on Algorithms for Planning and Control of Robot Motion.


Lee Cronk

Professor and Graduate Program Director

Department of Anthropology and Rutgers Center for Human Evolutionary Studies (CHES)



My main interest is in the role of culture in the evolution of human behavior. I use a framework based on animal signaling theory to explore such topics as mate choice, cooperation, the relationship between culture and behavior, and cross-cultural differences and similarities in perception and aesthetic judgments. I recently assembled a team of behavioral ecologists and computer scientists to study dance as a courtship signal, using motion capture animation to separate dancers’ movements from their outward appearance.

Lee Cronk received his B.A. in 1982 and his Ph.D. in 1989, from Northwestern University. He arrived at Rutgers in 1999 after having taught for ten years at Texas A&M University. He has conducted fieldwork in Kenya, Honduras, and Jamaica. He is the author of: That Complex Whole: Culture and the Evolution of Human Behavior (Westview, 1999) and From Mukogodo to Maasai: Ethnicity and Cultural Change in Kenya (Westview, 2004) and co-editor of Adaptation and Human Behavior: An Anthropological Perspective (Aldine de Gruyter, 2000). His work has been supported by the National Science Foundation, the Population Council, a Fulbright grant, and Rutgers’ Center for Human Evolutionary Studies.


Kristin Dana

Associate Professor

Department of Electrical and Computer Engineering


My primary research interests are computational skin models, imaging device design and prototyping, surface geometry, texture recognition, and the Rutgers Skin Texture Database and Columbia-Utrecht Reflectance and Texture Database

Kristin Dana received her Ph.D. in Electrical Engineering from Columbia University in 1999. Her awards include: National Science Foundation CAREER Award 2001; Best Presentation in Session Award, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1997; Sarnoff Technical Achievement Award 1994; NYU Computer Science and Engineering Award 1990; and General Electric "Faculty of the Future" Fellowship 1990.



Doug DeCarlo

Associate Professor

Department of Computer Science; Center for Cognitive Science


My main interests are in the cognitive science of visual interaction: I explore how accounts of human perception and communication can inform computer systems that engage in natural and effective visual presentation. My recent projects in computer graphics have focused on shape depiction in line drawings, meaningful abstraction in images, and conversational animation.

Doug DeCarlo received BS degrees in computer science and computer engineering from Carnegie Mellon in 1991, and his PhD in computer science from the University of Pennsylvania in 1998. He was a visiting fellow in the Department of Computer Science at Princeton University in 2002. He co-chaired NPAR 2006, the International Symposium on Non-Photorealistic Animation and Rendering, has served as a member of the program committees of major conferences in graphics and vision, and is an associate editor of ACM Transactions on Graphics. His research is supported by grants from the NSF.


Ahmed Elgammal

Assistant Professor,

Department of Computer Science



My primary research interest is computer vision and machine learning. My research focus includes human activity recognition, human motion analysis, tracking, human identification, and statistical methods for computer vision. I develop robust real-time algorithms to solve computer vision problems in areas such as visual surveillance, visual human-computer interaction, virtual reality, and multimedia applications. My interests also include research on document image analysis.

Dr. Elgammal received his M.Sc. degree in computer science and automatic control from University of Alexandria, Egypt in 1996. He received his Ph.D. degree in computer science from the University of Maryland, College Park, in 2002. He joined Rutgers in 2002 and is also a member of the Center for Computational Biomedicine Imaging and Modeling (CBIM) and the Center for Advanced Information Processing (CAIP) at Rutgers. Dr. Elgammal received the National Science Foundation CAREER Award in 2006.


Jacob Feldman


Department of Psychology; Center for Cognitive Science



My research interests lie primarily in perceptual organization, shape representation, and concept learning. In perceptual organization, I am especially interested in contour representation and extraction, the interaction of local and global cues in the formation of visual objects, and Bayesian models of image interpretation. In shape, I am especially interested in skeletal models of shape, decomposition into perceptual parts, and shape classification. In concept learning, I am especially interested in complexity-minimization approaches to generalization and model selection.

Jacob Feldman received his A.B. from Harvard College, 1986, from the Department of Psychology and Social Relations; and his Ph.D. in 1992 from the Department of Brain and Cognitive Science, M.I.T. He has been at Rutgers since 1992. He is on the editorial board of Cognition. His awards include: National Academy of Science Troland Research Award (2005); American Psychological Association George Miller award (2002); National Science Foundation “CAREER” award.



Pernille Hemmer


Department of Psychology


My research seeks to elucidate the relationship between mental representations and naturalistic environments. While my work has been focused on episodic and semantic memory, the overarching theme of my research is decision making in naturalistic environments. I use ecologically valid stimuli to capitalize on the idea that humans work in concert with their environment and that people use their knowledge and expectations to make decisions in a broad range of cognitive tasks. Specifically, I focus on complex environments in which people make real world decisions about situations where knowledge of the environment can be brought to bear. In these environments, I apply computational and Bayesian modeling to behavioral experiments. I explore how people use information from these environments in retrieving information from memory, and decision making in general.

Pernille Hemmer received her Ph.D. from the Department of Cognitive Science at the University of California, Irvine. She completed a post-doctoral fellowship in the Department of Psychology at Syracuse University before joining the faculty at Rutgers University in 2012. Her research has been recognized with a best-paper award for modeling of higher-level cognition.


Eileen Kowler


Department of Psychology; Center for Cognitive Science; Biomedical Engineering



Eye movements are an integral part of our interactions with the visual world. Tasks, such as reading, searching for objects, inspecting the contents of a visual scene, or navigating through the environment, require that we bring the eye quickly and accurately to important and useful locations. Remarkably, eye movements accomplish this goal with virtually no overt effort or awareness. The research in my lab is devoted to understanding how eye movements are planned, how they are carried out, and how we maintain the percept of a clear, stable and coherent world despite the continual changes in the visual array that eye movements produce. Our work emphasizes the active integration of eye movement planning with ongoing visual and cognitive processes, such as attention, perception and memory.

Eileen Kowler received her Ph.D. in Psychology from the University of Maryland in 1978, and joined the Psychology faculty at Rutgers in 1980 after postdoctoral work at New York University. She edited the reference work Eye Movements and Their Role in Visual and Cognitive Processes (published by Elsevier), and served as Section Editor for Behavioral Physiology and Visuomotor Control of the journal Vision Research from 1995 to 2004. She has been on the editorial boards of the Journal of Vision and Cognitive Brain Research. Her laboratory has been supported by grants from the Air Force Office of Scientific Research and NIH.





Peter Meer


Department of Electrical and Computer Engineering




My research interest is in application of modern statistical methods to image understanding problems.

Peter Meer received the Dipl. Engn. degree from the Bucharest Polytechnic Institute, Romania in 1971, and the D.Sc. degree from the Technion, Israel Institute of Technology, Haifa, in 1986, both in electrical engineering. From 1971 to 1979 he was with the Computer Research Institute, Cluj, Romania. From 1986-1990 he was Assistant Research Scientist at the Center for Automation Research, University of Maryland at College Park. In 1991 he joined Rutgers University. He has held visiting appointments in Japan, Korea, Sweden, Israel and France, and was on the organizing committees of numerous international workshops and conferences. He was an Associate Editor of the IEEE Transaction on Pattern Analysis and Machine Intelligence between 1998 and 2002, and was a member of the Editorial Board of Pattern Recognition between 1989 and 2005. He received best student paper in the 1999, and best paper in the 2000, IEEE Conference on Computer Vision and Pattern Recognition.


Dimitris Metaxas


Department of Computer Science; Biomedical Engineering

Director, Center for Computational Biomedicine, Imaging and Modeling (CBIM)



My lab is in the Center for Computational Biomedicine, Imaging and Modeling, which I direct. Our work focuses on physics-based techniques for the modeling, estimation, and synthesis of the shape and motion of human organs, limbs, and behavior. Current projects include: heart motion modeling, segmentation methods, minimally-invasive cancer biopsy procedures, cancer detection, gait modeling and correction, and genetic effects on sleeping patterns.

Dimitris Metaxas has been at Rutgers since 2001. From 1992 to 2001 he was in the Computer and Information Science Department of the University of Pennsylvania and Director of the VAST Lab. He received a Ph.D. in Computer Science from the University of Toronto, Ontario, Canada in 1992. Dr. Metaxas wrote Physics-based deformable models: Applications to computer vision, graphics and medical imaging (Kluwer Academic Press). He is on the Editorial Board of Medical Imaging, as Associate Editor of GMOD. He was awarded a Fulbright Fellowship in 1986, is a recipient of an NSF Research Initiation and Career awards, an ONR YIP, and is a Fellow of the American Institute of Medical and Biological Engineers, ACM and IEEE. His research has been funded by NSF, NIH, ONR, AFOSR and the ARO.


Melchi Michel


Department of Psychology



I study the human visual system with a focus on investigating how we integrate sensory information to make perceptual judgments, how we exploit statistical regularities in the environment, and how we adapt when these statistical regularities are altered or when new statistical contingencies are introduced. My research addresses a broad spectrum of topics ranging from basic questions about visual signal detection, visual search, perceptual learning, and neural coding to potential clinical applications in assessing the visual performance of retinopathic patients and radiologists and evaluating the potential impact of training interventions. Central to my approach are the treatment of vision as a problem of probabilistic inference and the use of ideal observers as a standard against which to compare human performance.

Melchi Michel received his Ph.D. from the Department of Brain and Cognitive Sciences at the University of Rochester in 2007. He was a postdoctoral fellow in the Center for Perceptual Systems at the University of Texas at Austin from 2007-2012 and joined the Rutgers faculty in 2012.




Thomas Papathomas


Department of Biomedical Engineering

Associate Director of the Laboratory of Vision Research




My laboratory investigates how the brain processes visual and auditory stimuli. Studies mainly involve psychophysical experiments with human observers. Current projects include the role of cognitive mechanisms in the recovery of three-dimensional shape from monocular or binocular 2-D retinal projections, interactions of auditory and visual processes, the role of attention, the study of excitatory and inhibitory neural influences on mechanisms of motion perception, and the development of computational models for the extraction of motion.

Thomas Papathomas received his BS, MS, and PhD (1977) from Columbia University from the Department of Electrical Engineering and Computer Science. He joined AT&T Bell Laboratories in 1977 and was in the Visual Perception Research Department, headed by Bela Julesz, from 1983 to 1989. He came to Rutgers in 1989. He edited Early Vision and Beyond, a volume of interdisciplinary research in psychophysics, neurophysiology, and computational vision (MIT Press, 1995). He has designed several exhibits in science museums, and his work was exhibited at the New York Arts Biennial in 1997. He has been a member of the Editorial Board of the International Journal of Imaging Systems and Technology. He holds 3 U.S. patents, and won the Best Paper Award of the IEEE Transactions on Industrial Electronics in 1986.



Vladimir Pavlovic

Assistant Professor

Department of Computer Science; Center for Cognitive Science; BIOMAPPS Institute for Quantitative Biology

Vladimir Pavlovic carries out research on stochastic modeling, segmentation and clustering of sequential data in computer vision and bioinformatics. He was among pioneers in developing the principles of probabilistic graphical models for learning dynamic systems from data, showing the benefits of dynamic Bayesian networks for elucidation of genomic structures and recognition and synthesis of the human gait. His Sequence Analysis and Modeling (SEQAM) Laboratory's current work focuses on the study of large-scale sequence clustering and classification methods for analysis of human motion and identification of biological species via DNA barcoding.

Vladimir Pavlovic received the Ph.D. in electrical and computer engineering from the University of Illinois at Urbana-Champaign in 1999. He holds a MS in electrical engineering from the University of Illinois in Chicago (1993) and the diploma in electrical engineering from the University of Novi Sad, Yugoslavia (1991). From 1999 until 2001 he was a research scientist at the Compaq's Cambridge Research Laboratories. He held a research faculty position at the Bioinformatics Program at Boston University in 2001-2002. He received the Distinguished Student Paper award at the 2005 International Conference on Machine Learning. He will be the Special Sessions and Panels Chair at the IEEE International Conference on Face and Gestures, Amsterdam, 2008. His research is currently funded by the National Science Foundation and the Department of Homeland Security.



Zenon Pylyshyn

Board of Governors Professor of Cognitive Science

Department of Psychology; Center for Cognitive Science



For the past fifteen years, Pylyshyn's research has dealt with the theoretical analysis of the nature of the human cognitive system that enables humans to perceive the world, as well as to reason and imagine. On the experimental side Pylyshyn has been concerned with exploring his Visual Indexing Theory (sometimes called the FINST theory), dealing with how human visual attention is allocated and how humans cognize objects and space.

Zenon Pylyshyn received a Ph.D. in Experimental Psychology from the University of Saskatchewan in 1963. He spent two years as a Canada Council Senior fellow and then joined the faculty at the University of Western Ontario in London. In 1994 he joined the faculty of Rutgers University. Pylyshyn is recipient of numerous fellowships and awards, including the Donald O. Hebb Award from the Canadian Psychological Association. In 1998 he was elected Fellow of the Royal Society of Canada. In 2004 he was awarded the Jean Nicod Prize in Paris and delivered the Jean Nicod lectures. He is past president of two international societies: the Society for Philosophy and Psychology, and the Cognitive Science Society. He is on the editorial boards of eight scientific journals and has been on several industrial or academic scientific advisory boards. He is author of Seeing and Visualizing: It's not what you think (MIT Press, 2004) and Computation and Cognition: Toward a Foundation for Cognitive Science (MIT Press, 1984).



Maggie Shiffrar


Department of Psychology - Newark Campus

973-353-5440 x235, x258


The goal of the research in my lab is to understand how the visual system interprets moving objects. We examine the relationships between visual physiology and visual perception for both "high" and "low" levels of analysis. This includes behavioral studies of the visual analysis of human movement, implicit memory of objects in motion, and the role of image segmentation cues in motion coherence, and visual memory for shape. At present, our primary focus of research is how visual experience, motor experience, and social processes all contribute to the visual analysis of human movement.

Maggie Shiffrar received her doctorate from Stanford University in 1990 and held Postdoctoral Fellowships at the Universite de Paris, France and NASA Ames Research Center. She was a Research Associate in the Movement and Perception Laboratory, CNRS & the University of the Mediterranean in 1998. She is on the editorial board of Cognitive Science and was previously a Consulting Editor of the Journal of Experimental Psychology: Human Perception and Performance. Awards include: Fellow of the American Psychological Association, 1993; Lansdowne Scholar Award, University of Vicptoria, 2003; Max Planck Gesellschaft Scholarship, 2003. Her research has been supported by grants from NEI, NSF, NASA, NATO, and NAAR.



Manish Singh

Associate Professor

Department of Psychology; Center for Cognitive Science



My research is concerned with the visual perception of objects and surfaces. I am interested in how the human visual system represents geometric information, and what implications this representation has for perception and cognition more generally. Specific topics include the visual representation of the shape of complex objects, and visual shape completion: how shape is completed or "filled in" when an object's boundary is partly occluded or camouflaged. I am also interested in how the brain generates "layered" representations of surfaces - in contexts involving partial occlusion and transparency - where two surfaces are represented along a single line of sight, one extending beneath the other.

Manish Singh received his Ph.D. from the Department of Cognitive Sciences at the University of California, Irvine, in 1998. He was a post-doctoral fellow in the Department of Brain & Cognitive Sciences at M.I.T. from 1998-2001. He joined Rutgers in 2001. His research has been supported by grants from the National Science Foundation.


Matthew Stone

Associate Professor

Department of Computer Science; Center for Cognitive Science




I work to develop computational models of face-to-face conversation. My overall strategy is based on modeling communicative intentions - how can we represent each utterance in terms of how its speaker intends to use it. This approach is particularly suited to multi-modal interactive communication, including the movements of the eyes, face, hands and body we use with speech in face-to-face conversation. My research builds on insights from Artificial Intelligence and Cognitive Science to bridge specialized models of linguistic representations and processes with general theories of perception, action, inference, and collaboration. We sometimes acquire and analyze people's utterances and conversations with one another, and other times develop proposals to account for linguistic intuitions, implement computer models, or carry out human-subjects evaluation of specific stimuli and artifacts.

Matthew Stone received his Ph.D. in 1998 in Computer and Information Science from the University of Pennsylvania. He was a postdoctoral fellow at Rutgers from 1998-1999 and joined the faculty in 1999. From 2005-2006 he was a Visiting Fellow in the School of Informatics at the University of Edinburgh. He currently serves on the editorial board of Computational Linguistics and is program co-chair for the 2007 North American Association for Computational Linguistics Human Language Technology Conference (NAACL HLT). His research is funded by NSF.


Karin Stromswold


Department of Psychology; Center for Cognitive Science




My research seeks to address the question: What makes human language special? Despite the seemingly intractable learnability problem posed by language acquisition, why do most children acquire language with ease? And despite the daunting computational problem posed by language processing, why do people process most sentences with ease? My research tackles these and other questions using a wide range of approaches including (1) cross-linguistic studies of typical and atypical language development to tease apart the aspects of language that must be learned from those that appear to be innate; (2) studies of sentence processing to investigate the factors that guide adult's and children's sentence processing; (3) behavior genetic studies of language to investigate the role that genetic (i.e., innate) factors play in language acquisition and processing; and (4) studies of children with perinatal risk factors to investigate how early biological environmental factors affect linguistic and nonlinguistic development.

Karin Stromswold received her PH.D. in Cognitive Science from M.I.T in 1990 and an M.D. from Harvard Medical School in General Medicine, in 1991. She is currently on the editorial boards of 4 journals.


Elizabeth Torres

Assistant Professor

Department of Psychology; Center for Cognitive Science; CBIM

732-445-6823 / 2469


My research looks at movements as another form of sensory input transduced and transformed into percepts that inform our Central and Peripheral Nervous Systems about our body in space and in time. We use the statistical signatures of movements to bridge body and mental processes in closed loop. This coupling facilitates co-adapting such processes with the environment and objectively quantifying learning gains. Our lab studies movements that range from intended to automatic. In particular we use spontaneous gestures as a gateway into the sensory capabilities and preferences of developing children to identify and to stimulate their best predispositions to learn. These include children with atypical developmental trajectories such as those with autism spectrum disorders who have been labeled “non verbal” and “low functioning”. These children can also spontaneously learn and acquire the intention to communicate with others through gestures and through various sensory modalities of their preference -a new finding in the lab that we are using to build communicative devices for non-verbal children as well as to diagnose and to treat different variants of autism. In collaboration with experts we are beginning exploration of the genetics underpinnings of autism.

Elizabeth Torres received her PhD from the Cognitive Science Department at the University of California San Diego in 2001, fully funded by an NIH fellowship. She then went on to receive postdoctoral training in electrophysiology as a Sloan-Swartz Fellow in Computational Neural Systems at the California Institute of Technology and as a Research Associate until she joined Rutgers in 2008. She serves as Associate Editor of Frontiers. Her research is funded by the NSF and the NJ Governor’s Council to treat autism.