Our paper "POLAR-Express: Efficient and Precise Formal Reachability Analysis of Neural-Network Controlled Systems" has been accepted for publication at IEEE TCAD. Congratulations to Weichao, Jiameng and our collaborators at Northwestern University, University of New Mexico, and University of Liverpool. Link to Tool: https://github.com/ChaoHuang2018/POLAR_Tool
Our team (with SRI International and Stony Brook) has been selected for Phase 2 of the TrojAI program. We will develop new technologies to defend AI systems from Trojan attacks.
New project sponsored by Intuit on developing a unified attribution analysis framework for interpretable deep learning.
Wenchao will be chairing the TPC subcommittee on Autonomous Systems (Automotive, Robotics, Drones) at DAC 2023
Please consider sending your strong papers to this track: https://www.dac.com/Conference/2023-Call-for-Contributions
Congratulations to Jiameng and our wonderful collaborators. POLAR substantially outperforms current SOTA techniques for reachability analysis of neural network-controlled systems. arXiv preprint: https://arxiv.org/abs/2106.13867