4th AFOSR Monterey Training Workshop on Computational Issues in Nonlinear Control
Topics at the Intersection of Deep Learning and Computational Nonlinear Control
The Embassy Suites by Hilton Monterey Bay Seaside
1441 Canyon Del Rey Blvd, Seaside, CA 93955
May 22-24, 2023
You can access video recordings of the presentations in the technical program. Simply click on the presentation title to start watching.
Welcome to the 4th AFOSR Monterey Workshop on Computational Nonlinear Control. We are delighted to bring together experts in the fields of deep learning, scientific computation, and control theory to exchange ideas on cutting-edge research at the intersection of these disciplines. The technical program will feature presentations on a diverse range of topics, including reinforcement learning, reachability and controllability, control of infinite dimensional systems, neural ODEs, optimization and system theory for deep learning, deep learning with applications to optimal control and MPC, and nonlinear filtering and estimation.
Please note that all presentations will be held in person, and on-site seating has already been fully booked. However, we will provide a zoom link for those who wish to attend the talks virtually. Thank you for your interest in our workshop, and we look forward to seeing you there!
Sponsor: Air Force Office of Scientific Research (AFOSR)
University of California, Davis
Organizing Committee
Wei Kang
Naval Postgraduate School and University of California, Santa Cruz
Arthur J. Krener
University of California, Davis and Naval Postgraduate School
Qi Gong
University of California, Santa Cruz
Zoom links: please download the file located at https://drive.google.com/file/d/1yL-nGmxYY9iTLZyuk1cz4JPFWnl11LcU/view?usp=sharing
Abstracts: https://drive.google.com/file/d/11nw6Yp2Tji7rQzdLSPY1Z8AvROq3mlij/view?usp=sharing
TECHNICAL PROGRAM
Monday, May 22, 2023
8:15-8:30 Fariba Fahroo and Fred Leve, AFOSR
8:30-9:00 Policy optimization and learning for optimal control with guarantees of robustness
Tamer Başar, UIUC
9:00-9:30 Stochastic nonlinear control via finite-dimensional spectral dynamic embedding
Na Li, Harvard
9:30-10:00 Data-driven representation-based control
Bo Dai, Georgia Tech
10:00-10:30 Coffee Break
10:30-11:00 Use and abuse of machine learning in scientific discovery
Daniel Tartakovsky, Stanford
11:00-11:30 Approximating reachable sets using learning-based methods
Claire Tomlin, UCB
11:30-1200 Data-driven reachability analysis for Gaussian process state space models
Paul Griffioen, UCB
12:00-1:30 Lunch
1:30-2:00 Modeling and digital twins for optimal control and design
John Burns, Virginia Tech
2:00-2:30 Neural operators for provably bypassing gain and control computations in PDE backstepping
Miroslav Krstic, UCSD
2:30-3:00 PDE-constrained models with neural network terms: optimization and global convergence
Konstantinos Spiliopoulos, BU
3:00-3:30 Coffee Break
3:30-4:00 Simulation-free generative modeling with neural ODEs
Ricky Chen, Meta
4:00-4:30 Controllability of neural differential equations
Enrique Zuazua, FAU
4:30-5:00 Certified and robust forward invariance in neural ODEs
Yuanyuan Shi, UCSD
5:00-5:15 Break
5:15-5:45 Remembrances of Marc Q Jacobs and Roger W. Brockett
Tuesday, May 23, 2023
8:15-8:30 Fariba Fahroo and Fred Leve, AFOSR
8:30-9:00 Errors and disturbances in learning dynamics
Eduardo Sontag, NEU
9:00-9:30 The role of systems theory in control oriented learning
Mario Sznaier, NEU
9:30-10:00 Leveraging multi-time Hamilton-Jacobi PDEs for certain scientific machine learning problems
Jerome Darbon, Brown
10:00-10:30 Coffee Break
10:30-11:00 Learning-based control for uncertain complex systems
Zhong-Ping Jiang, NYU
11:00-11:30 A Hamilton-Jacobi theory for optimal optimization
Isaac M Ross, NPS
11:30-1200 Stat-quad representations and dimensional reduction for optimal control problems
William McEneaney, UCSD
12:00-1:30 Lunch
1:30-2:00 Enhanced sampling for closed-loop optimal control design with deep neural networks
Jiequn Han, Flatiron Institute
2:00-2:30 Benchmark problems for learning-based optimal control
Tenavi Nakamura-Zimmerer, NASA
2:30-3:00 Upper bounds of neural network complexity with applications to control systems
Wei Kang and Qi Gong, NPS and UCSC
3:00-3:30 Coffee Break
3:30-4:00 Data completed nonlinear process models for optimization and control
James Rawlings, UCSB
4:00-4:30 Stability, Safety, and On-line Learning for Gaussian Process-Based Predictive Control
Rolf Findeisen, TU Darmstadt
4:30-5:00 Learning high-dimensional feedback laws for collective dynamics control
Dante Kalise, Imperial College
5:00-5:15 Break
5:15-5:45 Remembrances of Marc Q Jacobs and Roger W. Brockett
Wednesday, May 24, 2023
8:15-8:30 Fariba Fahroo and Fred Leve, AFOSR
8:30-9:00 Improved KKL observers via machine learning
Arthur J. Krener, NPS and UCD
9:00-9:30 Deep filtering: A computational approach
George Yin, UConn
9:30-10:00 Data-driven MPC for nonlinear systems with guarantees
Frank Allgöwer, Univ. of Stuttgart
10:00-10:30 Coffee Break
10:30-11:00 A neural network approach for real-time high-dimensional optimal control
Samy Wu Fung, Colorado School of Mines
11:00-11:30 Curse-of-dimensionality-free approximation of optimal value functions with neural networks
Mario Sperl, Univ. of Bayreuth
John S. Baras, UMD