Susmit Jha
Susmit Jha
Director, Neuro-symbolic Computing and Intelligence at SRI International
Dr. Susmit Jha is a Technical Director in Computer Science Laboratory at SRI, where he leads the research group on Neuro-symbolic Computing and Intelligence. He did his Ph.D. in Computer Science from UC Berkeley in 2011, where his thesis work on “Automated Synthesis Using Structurally Constrained Induction and Deduction” was awarded the Leon O Chua Award. His program synthesis work influenced the development of the FlashFill feature in Excel. Dr. Jha was at Intel Labs and Raytheon Technologies Research Center at Berkeley before joining SRI, where he received a Division Recognition Award in 2012 and Research Technology Scoping Award in 2014. He is the recipient of the 10 year Most Influential Paper award at IEEE/ACM ICSE 2020 and has published over 70 peer-reviewed publications with over 3100 citations in AI, ML and automated reasoning venues such as NeurIPS, ICLR, ICML, CVPR, AAAI, IJCAI, JAR, and CAV. Dr. Jha has been a Principal Investigator on DoD and US Govt. programs on trustworthy and neuro-symbolic AI, including DARPA Assured Autonomy, DARPA Symbiotic Design of Cyber-physical Systems, IARPA TrojAI, NSF Self-Improving Cyber-Physical, and Army Research Laboratory’s Internet of Battlefield Things REIGN CRA.
Track: Defense
Neuro-symbolic Learning with Large Foundation Models
State-of-the-art artificial intelligence (AI) and machine learning approaches have achieved remarkable success over the last decade and reached near-human level performance in many applications, such as natural language text generation and image processing. Despite these successes, these approaches are known to be fragile and not trustworthy, impeding their adoption in high-assurance safety-critical applications such as ISR. Phenomena such as hallucinatory fact generation in the large foundation models demonstrate that this fragility is persistent even with the scaling of the models. In this talk, we will describe a collaborative effort between SRI International and ACI/USMA, WestPoint, to develop a neuro-symbolic machine learning and reasoning approach inspired by Hierarchical Predictive Processing (HPP, a theory of mind) that uses context-based learning and decision-making to address the trinity of challenges – trustworthiness, robustness and interpretability of AI.
