Resilient Control of Cyber-Physical Systems with Distributed Learning - January 2020
PI(s) and Co-PI(s): Sayan Mitra and Geir Dullerud and Sanjay Shakkotai (U. Texas at Austin)
Researchers: Pulkit Katdare and Negin Musavi
HARD PROBLEM(S) ADDRESSED
This refers to Hard Problems, released November 2012.
Resiliency: Effective verification of safety and security properties of autonomous and cyber-physical systems
Metrics: How much data is necessary to achieve a certain level of confidence regarding a safety/security claim
PUBLICATIONS
Papers written as a result of your research from the current quarter only.
Working paper:
Optimistic Optimization for Statistical Model Checking with Regret Bounds, Musavi, Sun, Mitra, Shakkottai, and Dullerud, Jan 2020. Available online from https://arxiv.org/abs/1911.01537
Related recent publications:
Verifying PCTL Specifications on Markov Decision Processes via Reinforcement Learning, Yu Wang, Nima Roohi, Matthew West, Mahesh Viswanathan and Geir Dullerud, submitted.
Convergence Guarantees of Policy Optimization Methods for Markovian Jump Linear Systems, Joao Porto, Bin Hu, and Geir Dullerud, submitted.
Policy Learning of MDPs with Mixed Continuous/Discrete Variables: A Case Study on Model-Free Control of Markovian Jump Systems, Joao Porto, Bin Hu, and Geir Dullerud, submitted.
The Gossiping Insert-Eliminate Algorithm for Multi-Agent Bandits, Ronshee Chawla, Abishek Sankararaman, Ayalvadi J. Ganesh and Sanjay Shakkottai, accepted to appear in AISTATS 2020, Palermo, Sicily, June 2020.
Data-driven safety verification of complex cyber-physical systems, Chuchu Fan and Sayan Mitra. A chapter in the book titled Design Automation for Cyber-Physical Systems, edited by Mohammad Abdullah Al Faruquqe and Arquimedes Canedo, pages 107-143, Springer, 2019.
Using symmetry transformations in equivariant dynamical systems for their safety verification Hussein Sibai, Navid Mokhlesi and Sayan Mitra; accepted for publication in the proceedings of the Seventeenth International Symposium on Automated Technology for Verification and Analysis (ATVA), October 28-31, 2019, Taipei City, Taiwan. Nominated for best paper award
KEY HIGHLIGHTS
Each effort should submit one or two specific highlights. Each item should include a paragraph or two along with a citation if available. Write as if for the general reader of IEEE S&P.
The purpose of the highlights is to give our immediate sponsors a body of evidence that the funding they are providing (in the framework of the SoS lablet model) is delivering results that "more than justify" the investment they are making.
We organized and hosted the SOS Quaterly meeting for Fall 2019 in Chicago, November 5-6th, Discovery Partners Institute https://cps-vo.org/SoSLmtg/UIUC/2019
Optimal data rate estimation and model detection for safe autonomy. USC-MHI Cyber-Physical Systems Seminar promoted by the Center for Cyber-Physical Systems and the Internet-of-Things (CCI), lecture given by Sayan Mitra, October 16th, 2019.
Hyper-parameter Tuning for ML Models: A Monte-Carlo Tree Search (MCTS) Approach. USC-MHI Cyber-Physical Systems Seminar promoted by the Center for Cyber-Physical Systems and the Internet-of-Things (CCI), lecture given by Sanjay Shakkottai, October 2th, 2019.
Hyper-parameter Tuning for ML Models: A Monte-Carlo Tree Search (MCTS) Approach. ECE CSP Seminar, University of Michigan at Ann Arbor, lecture given by Sanjay Shakkottai, October 24, 2019.
Hyper-parameter Tuning for ML Models: A Monte-Carlo Tree Search (MCTS) Approach. EE Seminar, IIT Bombay, lecture given by Sanjay Shakkottai, December 17, 2019.
Hyper-parameter Tuning for ML Models: A Monte-Carlo Tree Search (MCTS) Approach. EE Seminar, Indian Institute of Science, lecture given by Sanjay Shakkottai, January 2, 2020.
Invited Speaker, "On the Throughput vs Accuracy Trade-Off for Streaming Unsupervised Classification", Sanjay Shakkottai, Workshop on Learning Theory 2, Tata Institute of Fundamental Research, January 3, 2020.
Invited Speaker, Formal Methods in Mathematics Workshop, Mathematics, Carnegie Mellon University, "Learning and Statistical Validation of Complex Cyber-PhysicalSystems", Geir Dullerud, January 7th, 2020 on logic in engineering systems.
CES Symposium Lecture, University of Texas, Austin, "Statistical Validation and Principle-Based Simulation of Complex Cyber-Controlled Systems", Geir Dullerud, December 3rd, 2019.
EDUCATIONAL ADVANCES
PI Mitra's new course Principles of Safe Autonomy at University of Illinois came to a successful conclusion in May. The course takes a deep dive into the seminal topics in object recognition, learning, localization, decision making, path planning, control, and safety verification. 25 students from ECE and CS are completed the course. The course team has designed 6 New programming assignments involving topics such as lane detection, road-sign recognition with deep neural networks, localization with particle filters, decision making with reinforcement learning, path planning with rapidly expanding random trees, and safety verification using simulation-driven proofs. The students used a high-fidelity, commercial-grade vehicle simulator (Righthook) for testing their programming assignments. Galois Inc. sponsored prizes for student projects. Find out more about the safe autonomy course and the student projects at https://publish.illinois.edu/safe-autonomy/