Weak Convergence Analysis of Online Neural Actor-Critic Algorithms

arXiv:2403.16825v1 Announce Type: new
Abstract: We prove that a single-layer neural network trained with the online actor critic algorithm converges in distribution to a random ordinary differential equation (ODE) as the number of hidden units and the number of training steps $\rightarrow \infty$. In the online actor-critic algorithm, the distribution of the data samples dynamically changes as the model is updated, which is a key challenge for any convergence analysis. We establish the geometric ergodicity of the data samples under …

abstract actor actor-critic algorithm algorithms analysis arxiv convergence cs.lg differential differential equation distribution equation hidden layer math.oc math.pr network neural network ordinary prove random stat.ml training type units