The Intel HPC Developer Conference at SC16 has announced its keynote speakers. Jonathan Cohen and Kai Li from Princeton will present, Going Where Neuroscience and Computer Science Have Not Gone Before.
Taking place Nov. 12-13 in Salt Lake City, the Intel HPC Developer Conference will bring together developers from around the world to discuss code modernization in high-performance computing.
Abstract:
“Neuroscience and computer science have benefited each other for decades. Important aspects of how the brain functions (e.g., learns to predict reward) have come from advances in computer science (e.g., reinforcement learning algorithms), and the recent advances in deep learning with back-propagation have their roots in discoveries made by cognitive neuroscientists several decades ago based on the development of artificial neural networks and their application to problems of learning and representation. Close interactions have also been central to methodological advances, for example the application of statistical analysis, machine learning algorithms and high-performance computing platforms to the analysis of neuroscientific data (such as firing pattern of populations of neurons, and the patterns of activity in human brain imaging data). However, current studies have only scratched the surface of the grand challenge of how human brains work. With over 100 billion neurons in the human brain, each of which makes contact with approximately 1,000 other neurons, it has been estimated that the number of potential circuits exceeds the number of molecules in the universe. The sheer magnitude of this problem poses an inexhaustible need for more sophisticated algorithms and powerful high-performance computing (HPC) platforms.”
Drs. Cohen and Li will discuss recent advances and opportunities for meeting this challenge, with a focus on the analysis of human brain imaging data. In particular, they will discuss a large research effort between Princeton and Intel Labs, attacking this challenge with machine learning and HPC methods. They will show how to analyze functional interactions of human brain regions from functional magnetic resonance imaging (fMRI) data, how to reduce computing time from years to minutes, and even real time. They will discuss some of the advances that are being made using such data analysis tools, including the ability to analyze mental states and functional interactions in real-time and how to build a real-time data analysis service for close-loop feedback trainings and potential brain diagnostics, using remote fMRI scanners.
Johnathan Cohen, Princeton Neuroscience Institute
Jonathan Cohen, Co-Director of Princeton Neuroscience Institute. Dr. Cohen has an M.D. from the University of Pennsylvania, did his residency training in psychiatry at Stanford University, and received a Ph.D. in cognitive and computational psychology from Carnegie Mellon University. His research focuses on the development of formal, mechanistically explicit models of the neural mechanisms responsible for cognitive control, and testing these in behavioral and neuroimaging studies in humans. He is one of the two founding Co-Directors of the Princeton Neuroscience Institute, is the author of over 250 scientific publications, is a fellow of the American Psychological Society and the American Association for the Advancement of Science, and a recipient of the American Psychological Association’s Distinguished Scientific Contribution Award.
Kai Li, Professor, Computer Science Department at Princeton University
Kai Li, Professor, Computer Science Department at Princeton University. Kai Li is a Paul M. Wythes ’55, P’86, and Marcia R. Wythes P’86 Professor at Princeton University, where he worked as a faculty member at Computer Science Department since 1986. His research expertise is in building parallel and distributed systems, deduplication storage systems, and data analysis and search for large datasets. He pioneered Distributed Shared Memory (DSM) that allow users to program using shared-memory programming model on a cluster of computers, which opened a new research area in parallel distributed systems. His work on user-level DMA on a cluster of computers evolved into the widely used RDMA standard in Infiniband. He co-led the ImageNet project, the largest publicly available knowledge base for computer vision, which propelled deep learning to become one of the most active research areas in machine learning. He co-founded Data Domain, Inc., in 2001, serving in roles as the initial CEO, CTO, and Chief Scientist. He is a Fellow of the ACM, and a Fellow of the IEEE, and a Member of the National Academy of Engineering.
Register for the Intel HPC Developer Conference
See our complete coverage of SC16, which takes place Nov. 13-18 in Salt Lake City.
