Ben Chasnov

PhD candidate at the University of Washington


Broadly speaking, my research interest is how optimizers interact with each other and with humans. More specifically, we study agents that use gradients to solve for the optimal actions given the (modeled or observed) actions of others. These settings have wide ranging applications in robust control and robust machine learning, human/machine interaction and computational neuroscience. My advisers are Sam Burden and Lillian Ratliff. I collaborate with the Autonomous Control Laboratory and the Computational Neuroscience Center community at UW.


My prior research at the Lab for Autonomous and Intelligent Robotics (LAIR) at Harvey Mudd College was on cooperative multi-agent underwater robotics and aerial vehicles, with an emphasis of deploying robot systems in harsh environments such as confined caves and open waters.

Working Papers

Gradient-Based Multi-Agent Learning with Time-Scale Separation

with Lillian J. Ratliff and Samuel A. Burden

Stability of Gradient Learning Dynamics in Continuous Games:

Vector Action Spaces

pre-print

with Dan Calderone, Behcet Acikmese, Samuel A. Burden, Lillian J. Ratliff

Human Autonomy Interface for Optimization-Based Control

with Daniel Sullivan, Margaraet Skye Mceowen, Oliver Sheridan, Michael Szmuk, Behcet Acikmese

Publications

Stability of Gradient Learning Dynamics in Continuous Games:

Scalar Action Spaces

with Dan Calderone, Behcet Acikmese, Samuel A. Burden, Lillian J. Ratliff

In IEEE Conference on Decision and Control (CDC), Dec 2020.

pdf

Implicit Learning Dynamics in Stackelberg Games:

Equilibria Characterization, Convergence Analysis, and Empirical Study

with Tanner Fiez and Lillian J. Ratliff

In Thirty-seventh International Conference on Machine Learning (ICML), July 2020.

pdf

Convergence Analysis of Gradient-Based Learning in Continuous Games

with Lillian J. Ratliff, Eric Mazumdar, Samuel A. Burden.

In Uncertainty in Artificial Intelligence (UAI), pp. 935-944. Proceedings of Machine Learning Research, 2019.

pdf

Experiments with Sensorimotor Games in Dynamic Human/Machine Interaction.

with Momona Yamagami, Behnoosh Parsa, Lillian J. Ratliff, Samuel A. Burden.

In Proceedings of SPIE Micro- and Nanotechnology Sensors, Systems, and Applications XI, May 2019.

doi

Towards Three-Dimensional Underwater Mapping Without Odometry.

with Alistair Dobke, Joshua Vasquez, Lauren Lieu, Christopher Clark, Ian Dunn, Zoe J. Wood, and Timothy Gambin.

In Unmanned Untethered Submersible Technology Conference 2013 Proceedings: Portsmouth, NH. 2013.

pdf

Workshop Papers

Opponent Anticipation via Conjectural Variations

with Tanner Fiez and Lillian J. Ratliff.

In NeurIPS Smooth Games Optimization and Machine Learning Workshop: Bridging Game Theory and Deep Learning, Dec 2019.

pdf

Finite-Time Convergence of Gradient-Based Learning in Continuous Games.

with Lillian J. Ratliff, Daniel Calderone, Eric Mazumdar, and Samuel A. Burden.

In AAAI Workshop on Reinforcement Learning in Games, Jan 2019.

pdf

Talks

I’ve given oral presentations at