Adaptive Trust Region Policy Optimization: Global Convergence and Faster Rates for Regularized MDPs

09/06/2019
by   Lior Shani, et al.
0

Trust region policy optimization (TRPO) is a popular and empirically successful policy search algorithm in Reinforcement Learning (RL) in which a surrogate problem, that restricts consecutive policies to be `close' to one another, is iteratively solved. Nevertheless, TRPO has been considered a heuristic algorithm inspired by Conservative Policy Iteration (CPI). We show that the adaptive scaling mechanism used in TRPO is in fact the natural "RL version" of traditional trust-region methods from convex analysis. We first analyze TRPO in the planning setting, in which we have access to the model and the entire state space. Then, we consider sample-based TRPO and establish Õ(1/√(N)) convergence rate to the global optimum. Importantly, the adaptive scaling mechanism allows us to analyze TRPO in regularized MDPs for which we prove fast rates of Õ(1/N), much like results in convex optimization. This is the first result in RL of better rates when regularizing the instantaneous cost or reward.

READ FULL TEXT
POST COMMENT

Comments

There are no comments yet.

Authors

page 1

page 2

page 3

page 4

10/26/2021

EnTRPO: Trust Region Policy Optimization Method with Entropy Regularization

Trust Region Policy Optimization (TRPO) is a popular and empirically suc...
05/20/2020

Mirror Descent Policy Optimization

We propose deep Reinforcement Learning (RL) algorithms inspired by mirro...
06/14/2020

Optimistic Distributionally Robust Policy Optimization

Trust Region Policy Optimization (TRPO) and Proximal Policy Optimization...
01/29/2019

Trust Region-Guided Proximal Policy Optimization

Model-free reinforcement learning relies heavily on a safe yet explorato...
09/24/2021

Regularization Guarantees Generalization in Bayesian Reinforcement Learning through Algorithmic Stability

In the Bayesian reinforcement learning (RL) setting, a prior distributio...
06/24/2019

Deep Conservative Policy Iteration

Conservative Policy Iteration (CPI) is a founding algorithm of Approxima...
03/07/2019

When random search is not enough: Sample-Efficient and Noise-Robust Blackbox Optimization of RL Policies

Interest in derivative-free optimization (DFO) and "evolutionary strateg...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.