Continuous Learning of System Security Thru Deep Topological Learning
|
ABSTRACT Continuous Learning of System Security Thru Deep Topological Learning We propose to combine improved behavioral description using Topological Data Analysis with recent advances in machine learning to further enhance the continuous learning of system-level security. Our goal is to improve detecting, clustering, classifying, and tracking of various patterns-of-life trajectories of system by providing the capability to better distinguish behavioral types at all scales. Slides found here. |
BIO
Professor Chin directs LISP (Learning, Intelligence + Signal Processing) group in the computer science department at Boston University, where he and his students are researching into fundamental questions such as “Can Intelligence be learned?” at the intersection of signal processing, machine learning, game theory, extremal graph theory, and computational neuroscience. He is also Senior Technical Director at BBN in Cambridge, MA. He has also held positions of Chief Scientist at Systems & Technology Research (STR), Chief Scientist – Decision Systems at Draper Laboratory, Before moving back to the New England in 2013, he was a co-director of DSP group in the Electrical and Computer Engineering (ECE) Department at Johns Hopkins University and a Chief Scientist in Cyber Technology Branch at Johns Hopkins Applied Physics Laboratory. He was a visiting fellow of London Institute of Mathematical Sciences and has held visiting positions at Tufts University (CS), Harvard University (Center of Mathematical Applications) and MIT (Dept of Brain and Cognitive Science). He’s currently an associate editor of IEEE Transactions on Computational Social Systems, and has served as conference co–chair of the annual SPIE/DSS Conference on Cyber Sensing, and symposium chair in GlobalSIP conference. Since completing his PhD for developing differential geometric methods to understand Einstein’s field equations, he has been passionate about developing geometric and topological methods to learn and understand information in general – signals (neural, RF, images, videos, hyper-spectral, etc.), graphs (social networks, communication networks, etc.) and human interactions via game theory. Most of his research is being (and has been) supported by NSF, NIH, AFOSR, DARPA, ODNI, ONR, OSD, and others and has been published at conferences such as NeurIPS, ICASSP, ISIT, etc. and journals such as Science Advances, IEEE Transactions and Journal of Machine Learning, etc. Peter is a Phi Beta Kappa graduate of Duke University where he was a triple major in computer science, math and electrical engineering. He received his PhD in mathematics from MIT.