Invited Speakers
John Donoghue, Brown University
http://brainsciences.brown.edu/departments/faculty/donoghue.html
From Mind to Movement: Developing Neurotechnologies to Restore Lost Function
Abstract: Neurotechnology is an exciting new field that has the potential to provide a variety of devices that can restore or possibly augment human brain function. The field has roots that go back at least to the middle of the last century when neuroscientists and surgeons demonstrated that the brain's activity could be read out and influenced through physical devices coupled directly to the nervous system. Recenty, there has been a marked increase in the pace of new developments in this emerging field. Neurotechnologies are now available for humans to treat deafness and relieve the symptoms of Parkinson's disease. More recent experiments in monkeys have shown that activity reflecting intended movement can be read out directly from populations of neurons in the motor cortex and this signal can be used as a substitute for hand motion to move a computer cursor. These advances stem from our increased knowledge of movement coding in the brain, from more powerful technology, and from mathematical tools that can decode the activity of populations at the "speed of thought". This proof of concept work has stimulated the development of a neuromotor prosthetic systems that may provide a new, fast and natural output signal for paralyzed humans. This system could be used to interact with the world through computers, robots, or to move their limbs through physical connections to their paralyzed muscles. These assistive technologies can improve functional independence and provide significant health benefits. Advances in neuromotor prosthetics could also enhance the development of other prosthetic devices for artificial vision and for products that could diagnose and treat disorders such as epilepsy or depression. The emergence of the ability to read out brain activity, or alter it, poses interesting new ethical questions that require careful consideration.
Bio: John Donoghue is a Professor and Chairman of the Neuroscience Department at Brown University. He received his PhD from Brown University in 1979 and was a postdoctoral fellow at NIH. His research interests include neural codes, motor control, and brain computer interfaces.
Gerd Gigerenzer, Max Planck Institute for Human Development
http://www.mpib-berlin.mpg.de/en/mitarbeiter/cv/gigerenzer.htm
Fast and Frugal Heuristics: The Adaptive Toolbox
Abstract: Heuristics are solutions to problems when optimization is out of reach, too costly, too slow, or socially unacceptable. The heuristics in the adaptive toolbox are anchored in the mind and the environment. They are embodied in the sense that they can exploit evolved capacities of the human mind (such as recognition memory), which allows judgments to be quick. They are anchored in the environment in the sense that they can exploit statistical or social structures (such as signal-noise ratio), which allows one to ignore information -- less can be more. The rationality of heuristics is ecological, not logical. The study of fast and frugal heuristics centers on two questions. What heuristics do people have at their disposal, and in what situations do they use a given heuristic? This descriptive question concerns the contents of homo sapiens adaptive toolbox: the building blocks of heuristics and the abilities they exploit. The second question is prescriptive. In what situations is the use of a heuristic reasonable? A heuristic is not good or bad, or rational or irrational per se, but is rational to the degree that it can exploit environmental structures. The study of the ecological rationality of a heuristic specifies, by means of analysis or simulation, the class of situations in which a heuristic is successful.
Bio: Gerd Gigerenzer is Director of the Center for Adaptive Behavior and Cognition at the Max Planck Institute for Human Development in Berlin and former Professor of Psychology at the University of Chicago. He won the AAAS Prize for the best article in the behavioral sciences. He is the author of "Calculated Risks: "How To Know When Numbers Deceive You," the German translation of which won the Scientific Book of the Year Prize in 2002. He has also published two academic books on heuristics, "Simple Heuristics That Make Us Smart" (with Peter Todd & The ABC Research Group) and "Bounded Rationality: The Adaptive Toolbox" (with Reinhard Selten, a Nobel laureate in economics).
Nati Linial, Hebrew University of Jerusalem
http://www.cs.huji.ac.il/~nati/
Expanders, Eigenvalues and All That
Abstract: A standard way of representing an n-vertex graph is through its (n x n, symmetric 0/1) adjacency matrix. Can we learn anything useful about the properties of the graph by considering the eigenvalues and eigenvectors of this matrix? This idea would initially seem too naive to be of any worth, but it turns out to be rather effective. The reason for this success may be that this is an exact discrete analogue of a very powerful process in modern analysis and geometry, namely, the spectral analysis of the Laplace operator. Specifically, the so-called Tutte matrix of a graph is an exact discrete analogue of the Laplace operator defined on manifolds. There are several natural graph parameters for which this approach works particularly well. In particular, there is a tight connection between the expansion properties of the graph and the spectral gap of its adjacency matrix. The first half of my talk will be a review of this fascinating field. In the second half, I will review recent work (with Yonatan Bilu) on the generation of (nearly)-optimal expander graphs. One key ingredient of this recent work provides a tighter connection between combinatorial properties of the graph (discrepancy) and the spectral gap of its matrix.
Bio: Born in Haifa, Israel at 1953. Undergraduate studies in Mathematics at the Technion, PhD in Mathematics from the Hebrew University 1978, with Micha Perles in the field of graph theory. Dr. Linial works mostly at the boundary between Mathematics and Computer Science. He gets the most excitement in research when he manages to introduce deep mathematical ideas into computer science, for example tools of harmonic analysis to analyze boolean functions, methods from Banach Space theory and metric spaces. Ideas from these fields have been used to develop new approximation algorithms and new approaches to the analysis of large data sets. Under his more applied hat he also work in bioinformatics.
Bernard Palsson, University of California, San Diego
http://gcrg.ucsd.edu/personnel/palsson.html
Systems Biology: Bringing Genomes to Life
Abstract: High throughput (HT) data generation in biology has led to the availability of vast amounts of chemical compositional data about cells. These developments have led to the emergence of systems biology that is widely viewed as being comprised of four steps: 1) information about cellular components; 2) reconstruction of biochemical reaction networks; 3) formulation of in silico model of network functions (i.e. phenotypes) and 4) measurement of phenotypic responses and their comparison to computed properties. Disagreement leads to an iterative model building procedure. HT phenotyping is one of the limiting steps in this process.
Reconstruction of genome-scale networks for metabolism and regulation in single cellular organisms in now possible, and efforts in reconstructing networks in human cells have begun. In silico models that characterize their function can be used to analyze, interpret and predict the genotype-phenotype relationship. Reconstructed genome-scale models for e. coli and yeast that include metabolism, regulation and transcription/translation have been formulated. These models integrate and represent a wide variety of high-throughput data.
Genome-scale models can be used to analyze the phenotypic consequences of gene deletions, optimal growth rates, the outcome of adaptive evolution, and for design of strains for bioprocessing. Examples in all these categories will be given, with emphasis on the computational and experimental analysis of adaptive evolution. Full characterization of adaptive evolutionary processes in terms of genome-wide expression profiling and full DNA re-sequencing has been performed. Thus both the genetic and epigenetic changes underlying adaptive evolution have been measured on a genome-scale and this data can be interpreted with the genome-scale in silico models.
Bio: Bernhard Palsson is a Professor of Bioengineering and Adjunct Professor of Medicine at the University of California, San Diego. Professor Palsson is the author of over 150 peer reviewed scientific articles and inventor of over 20 U.S. patents, many of which are in the area of hematopoietic stem cell transplantation, cell culture technology, bioreactor design, gene transfer, cell separations, gene finding, in silico model building and metabolic engineering. He holds a PhD from the University of Wisconsin that he earned in 1984. Professor Palsson held a faculty position at the University of Michigan for 11 years from 1984 to 1995. He received an Institute of International Education Fellowship in 1977, Rotary Fellowship in 1979, a NATO fellowship in 1984, was named the G.G. Brown Associate Professor at Michigan in 1989, a Fulbright Fellow in 1995, an Ib Henriksen Fellow in 1996, the Olaf Hougen Professorship at the University of Wisconsin in 1999, and the Lindbergh Tissue Engineering award in 2001. He sits on the editorial boards of several bioengineering and biotechnology journals. His current research at UCSD focuses on: 1) the reconstruction of genome-scale biochemical reaction networks; 2) the development of mathematical analysis procedures for genome-scale models; and 3) the experimental verification of genome-scale models with current emphasis on cellular metabolism and transcriptional regulation in e. coli and yeast. He co-founded a biotechnology company, AASTROM BIOSCIENCES (NASDAQ: ASTM) in 1988, where he served as the Vice President of Developmental Research for two years. Dr. Palsson is the founder and co-founder of ONCOSIS, a company that is focused on the purging of occult tumor cells in autologous bone marrow transplants, CYNTELLECT, that is focused on instrumentation for high-throughput screening and in situ cell sorting and processing, GENOMATICA, a company that is focused on in silico biology, and the Iceland Genomics Corporation a company that is focused on tracing the genetic basis for common human diseases in the Icelandic population.
Shimon Ullman, Weizmann Institute of Science
http://www.wisdom.weizmann.ac.il/~shimon/
Classification, Recognition and Segmentation Using Fragments Hierarchy
Abstract: The talk will describe a general approach to visual classification, recognition and segmentation. The approach is based on representing shapes within a class by a hierarchy of shared sub-structures called fragments. The fragments are sub-images selected automatically from a training set of images, by maximizing the mutual information of the fragments and the class they represent. For the task of individual recognition, these fragments are generalized to become extended fragments, which are equivalence sets of fragments, representing the same object part under different viewing conditions.
By a repeated application of the same feature extraction process, the classification fragments are broken down successively into their own optimal components. The resulting feature hierarchy is used to classify new images by the application of a feed-forward sweep from low to high levels of the hierarchy, followed by a sweep from the high to low levels.
Finally, image segmentation into an object and background is combined in this approach with the classification process. This is in contrast with the more common view, in which image segmentation is performed first, in a bottom-up manner, followed by object recognition. I will describe the approach, computational results, and possible relationships between the model and properties of parts of the primate visual system involved in object perception.
Bio: Shimon Ullman is the Ruth and Samy Cohn Professor of Computer Science in the Department of Computer Science and applied mathematics at the Weizmann Institute of Science, Rehovot, Israel. He received his BsC from the Hebrew University in Jerusalem, and PhD from MIT, where he has been a Professor in the Brain and Cognitive Science Department and in the Artificial Intelligence Laboratory. His main areas of research are human vision, computer vision, and brain modeling.