Counter-example guided synthesis of neural network Lyapunov functions for piecewise linear systems
We introduce an algorithm for synthesizing and verifying piecewise linear Lyapunov
functions to prove global exponential stability of piecewise linear dynamical systems. The …
functions to prove global exponential stability of piecewise linear dynamical systems. The …
Value iteration in continuous actions, states and time
Classical value iteration approaches are not applicable to environments with continuous
states and actions. For such environments, the states and actions are usually discretized …
states and actions. For such environments, the states and actions are usually discretized …
Reinforcement learning with non-exponential discounting
M Schultheis, CA Rothkopf… - Advances in neural …, 2022 - proceedings.neurips.cc
Commonly in reinforcement learning (RL), rewards are discounted over time using an
exponential function to model time preference, thereby bounding the expected long-term …
exponential function to model time preference, thereby bounding the expected long-term …
Hamilton-Jacobi deep Q-Learning for deterministic continuous-time systems with lipschitz continuous controls
In this paper, we propose Q-learning algorithms for continuous-time deterministic optimal
control problems with Lipschitz continuous controls. A new class of Hamilton-Jacobi …
control problems with Lipschitz continuous controls. A new class of Hamilton-Jacobi …
Neural network optimal feedback control with guaranteed local stability
T Nakamura-Zimmerer, Q Gong… - IEEE Open Journal of …, 2022 - ieeexplore.ieee.org
Recent research shows that supervised learning can be an effective tool for designing near-
optimal feedback controllers for high-dimensional nonlinear dynamic systems. But the …
optimal feedback controllers for high-dimensional nonlinear dynamic systems. But the …
Kernel-Based Optimal Control: An Infinitesimal Generator Approach
P Bevanda, N Hosichen, T Wittmann… - arXiv preprint arXiv …, 2024 - arxiv.org
This paper presents a novel approach for optimal control of nonlinear stochastic systems
using infinitesimal generator learning within infinite-dimensional reproducing kernel Hilbert …
using infinitesimal generator learning within infinite-dimensional reproducing kernel Hilbert …
Robust value iteration for continuous control tasks
When transferring a control policy from simulation to a physical system, the policy needs to
be robust to variations in the dynamics to perform well. Commonly, the optimal policy overfits …
be robust to variations in the dynamics to perform well. Commonly, the optimal policy overfits …
POMDPs in continuous time and discrete spaces
Many processes, such as discrete event systems in engineering or population dynamics in
biology, evolve in discrete space and continuous time. We consider the problem of optimal …
biology, evolve in discrete space and continuous time. We consider the problem of optimal …
Continuous-time fitted value iteration for robust policies
M Lutter - Inductive Biases in Machine Learning for Robotics and …, 2023 - Springer
One approach to obtain the optimal control inputs that maximize the reward, is to solve the
equation, as this differential equation expresses a sufficient and necessary condition for …
equation, as this differential equation expresses a sufficient and necessary condition for …
Neural optimal control using learned system dynamics
We study the problem of generating control laws for systems with unknown dynamics. Our
approach is to represent the controller and the value function with neural networks, and to …
approach is to represent the controller and the value function with neural networks, and to …