When Will Generative Adversarial Imitation Learning Algorithms Attain Global Convergence
arXiv:2006.13506
Abstract
Generative adversarial imitation learning (GAIL) is a popular inverse reinforcement learning approach for jointly optimizing policy and reward from expert trajectories. A primary question about GAIL is whether applying a certain policy gradient algorithm to GAIL attains a global minimizer (i.e., yields the expert policy), for which existing understanding is very limited. Such global convergence has been shown only for the linear (or linear-type) MDP and linear (or linearizable) reward. In this paper, we study GAIL under general MDP and for nonlinear reward function classes (as long as the objective function is strongly concave with respect to the reward parameter). We characterize the global convergence with a sublinear rate for a broad range of commonly used policy gradient algorithms, all of which are implemented in an alternating manner with stochastic gradient ascent for reward update, including projected policy gradient (PPG)-GAIL, Frank-Wolfe policy gradient (FWPG)-GAIL, trust region policy optimization (TRPO)-GAIL and natural policy gradient (NPG)-GAIL. This is the first systematic theoretical study of GAIL for global convergence.
Submitted for publication
References in corpus (8)
- Solving a Class of Non-Convex Min-Max Games Using Iterative First Order Methods
- Neural Proximal/Trust Region Policy Optimization Attains Globally Optimal Policy
- Global Convergence of Policy Gradient Methods to (Almost) Locally Optimal Policies
- Imitation Learning as -Divergence Minimization
- Reanalysis of Variance Reduced Temporal Difference Learning
- On the Global Convergence of Imitation Learning: A Case for Linear Quadratic Regulator
- On Computation and Generalization of Generative Adversarial Imitation Learning
- Generative Adversarial Imitation Learning with Neural Networks: Global Optimality and Convergence Rate