University of Massachusetts Amherst

Search Google Appliance


Talk: deepCode -- Peter Chin and Felipe Vilas-Boas (Draper Labs)

Wednesday, September 21, 2016 - 13:30 to 14:15

Title: deepCode
Peter Chin and Felipe Vilas-Boas (Draper Labs)

Abstract: Everyday, people (from neophyte CS 101 students to expert C++ programmers) are writing computer programs everywhere. And many of these programs are publicly available from sites such as Github, Gitlab, etc., where these codes are uploaded and shared. What can we possibly learn from this big data of SW projects? DARPA MUSE (Mining and Understanding of Software Enclaves) seeks to answer such questions, and I will describe Draper Laboratory’s approach for MUSE using a machine learning technique called deep learning.

Dr. Peter Chin is Chief Scientist – Decision Systems at the Charles Stark Draper Laboratory, where he conducts research in the areas of compressive sensing, machine learning, data fusion, extremal graph theory, game theory, multiple hypothesis tracking, quantum-game inspired cyber-security, sensor resource management, and computational neuroscince with grants from ONR, AFOSR, OSD, NSF, and DARPA. He is/was PI for numerous grants, including a 4-year ONR grant (Geometric Multi-Resolution Analysis) where he is applying hierarchical geometric sparse recovery techniques to high dimensional data with low intrinsic dimension, a 5-year grant from AFOSR (topological game-sematic cyber security) and a 3-year grant from OSD (Advanced Math BAA), where he is extending the traditional MHT theory for more general settings and applications. He is also currently the PI of Draper’s effort on DARPA/MUSE (Mining and Understanding of Software Enclaves) where he is developing deep learning approaches to find patterns of software vulnerabilities. He is also a research professor in the Dept. of Computer Science at Boston University where he directs LISP (Learning, Intelligence + Signal Processing) group ( and teaches in the field of machine learning and game theory. He is a Fellow of the Hariri Institute of Computing at BU as well as a visiting fellow at the London Institute of Mathematical Sciences. He is a Phi Beta Kappa graduate of Duke University where he was a triple major in EE, CS and Math and received his Ph.D. in Mathematics from MIT.

Felipe Vilas-Boas is a Research Engineer in the Cyber and Information Security group at Draper. Mr. Vilas-Boas has been at Draper for over 3 years and his primary research areas are computer security, embedded device programming and security, and trusted computing. He is involved in various of programs within Draper that range from static and dynamic analysis of software and firmware, applications of machine learning, and development of new and existing security features for existing hardware architectures. Prior to joining Draper in 2013, Mr. Vilas-Boas worked at BlueRISC, an Amherst, MA based security company, which specializes in providing a hardware root of trust device for new and existing systems. He was tasked to performed in various roles such as firmware and driver development of PCI-Express devices, application and software stack development for internal tools and runtime applications. He is a graduate of UMass Amherst.