Mean-Field Generalisation Bounds for Learning Controls in Stochastic Environments
By: Boris Baros, Samuel N. Cohen, Christoph Reisinger
Potential Business Impact:
Teaches computers to make smart choices from data.
We consider a data-driven formulation of the classical discrete-time stochastic control problem. Our approach exploits the natural structure of many such problems, in which significant portions of the system are uncontrolled. Employing the dynamic programming principle and the mean-field interpretation of single-hidden layer neural networks, we formulate the control problem as a series of infinite-dimensional minimisation problems. When regularised carefully, we provide practically verifiable assumptions for non-asymptotic bounds on the generalisation error achieved by the minimisers to this problem, thus ensuring stability in overparametrised settings, for controls learned using finitely many observations. We explore connections to the traditional noisy stochastic gradient descent algorithm, and subsequently show promising numerical results for some classic control problems.
Similar Papers
Robust mean-field control under common noise uncertainty
Optimization and Control
Helps groups of robots make better decisions together.
Towards optimal control of ensembles of discrete-time systems
Optimization and Control
Helps control many things at once.
Deep Learning for Continuous-time Stochastic Control with Jumps
Machine Learning (CS)
Teaches computers to make smart choices automatically.