skip to main content

Explore Research Products in the NSF-PAR

Title: Adversarially Trained Actor Critic for Offline Reinforcement Learning
We propose Adversarially Trained Actor Critic (ATAC), a new model-free algorithm for offline reinforcement learning (RL) under insufficient data coverage, based on the concept of relative pessimism. ATAC is designed as a two-player Stackelberg game: A policy actor competes against an adversarially trained value critic, who finds data-consistent scenarios where the actor is inferior to the data-collection behavior policy. We prove that, when the actor attains no regret in the two-player game, running ATAC produces a policy that provably 1) outperforms the behavior policy over a wide range of hyperparameters that control the degree of pessimism, and 2) competes with the best policy covered by data with appropriately chosen hyperparameters. Compared with existing works, notably our framework offers both theoretical guarantees for general function approximation and a deep RL implementation scalable to complex environments and large datasets. In the D4RL benchmark, ATAC consistently outperforms state-of-the-art offline RL algorithms on a range of continuous control tasks.  more » « less
Award ID(s):
2141781
NSF-PAR ID:
10394017
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
Proceedings of the 39th International Conference on Machine Learning
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ( , Advances in neural information processing systems)
    null (Ed.)
    Most prior approaches to offline reinforcement learning (RL) have taken an iterative actor-critic approach involving off-policy evaluation. In this paper we show that simply doing one step of constrained/regularized policy improvement using an on-policy Q estimate of the behavior policy performs surprisingly well. This one-step algorithm beats the previously reported results of iterative algorithms on a large portion of the D4RL benchmark. The simple one-step baseline achieves this strong performance without many of the tricks used by previously proposed iterative algorithms and is more robust to hyperparameters. We argue that the relatively poor performance of iterative approaches is a result of the high variance inherent in doing off-policy evaluation and magnified by the repeated optimization of policies against those high-variance estimates. In addition, we hypothesize that the strong performance of the one-step algorithm is due to a combination of favorable structure in the environment and behavior policy. 
    more » « less
  2. ; ; ; ( , Advances in neural information processing systems)
    Most prior approaches to offline reinforcement learning (RL) have taken an iterative actor-critic approach involving off-policy evaluation. In this paper we show that simply doing one step of constrained/regularized policy improvement using an on-policy Q estimate of the behavior policy performs surprisingly well. This onestep algorithm beats the previously reported results of iterative algorithms on a large portion of the D4RL benchmark. The one-step baseline achieves this strong performance while being notably simpler and more robust to hyperparameters than previously proposed iterative algorithms. We argue that the relatively poor performance of iterative approaches is a result of the high variance inherent in doing off-policy evaluation and magnified by the repeated optimization of policies against those estimates. In addition, we hypothesize that the strong performance of the one-step algorithm is due to a combination of favorable structure in the environment and behavior policy. 
    more » « less
  3. ; ; ( , Transportation Research Record: Journal of the Transportation Research Board)
    null (Ed.)
    Given the aging infrastructure and the anticipated growing number of highway work zones in the U.S.A., it is important to investigate work zone merge control, which is critical for improving work zone safety and capacity. This paper proposes and evaluates a novel highway work zone merge control strategy based on cooperative driving behavior enabled by artificial intelligence. The proposed method assumes that all vehicles are fully automated, connected, and cooperative. It inserts two metering zones in the open lane to make space for merging vehicles in the closed lane. In addition, each vehicle in the closed lane learns how to adjust its longitudinal position optimally to find a safe gap in the open lane using an off-policy soft actor critic reinforcement learning (RL) algorithm, considering its surrounding traffic conditions. The learning results are captured in convolutional neural networks and used to control individual vehicles in the testing phase. By adding the metering zones and taking the locations, speeds, and accelerations of surrounding vehicles into account, cooperation among vehicles is implicitly considered. This RL-based model is trained and evaluated using a microscopic traffic simulator. The results show that this cooperative RL-based merge control significantly outperforms popular strategies such as late merge and early merge in terms of both mobility and safety measures. It also performs better than a strategy assuming all vehicles are equipped with cooperative adaptive cruise control. 
    more » « less
  4. ; ; ; ; ( , The Tenth International Conference on Learning Representations)
    Existing offline reinforcement learning (RL) methods face a few major challenges, particularly the distributional shift between the learned policy and the behavior policy. Offline Meta-RL is emerging as a promising approach to address these challenges, aiming to learn an informative meta-policy from a collection of tasks. Nevertheless, as shown in our empirical studies, offline Meta-RL could be outperformed by offline single-task RL methods on tasks with good quality of datasets, indicating that a right balance has to be delicately calibrated between "exploring" the out-of-distribution state-actions by following the meta-policy and "exploiting" the offline dataset by staying close to the behavior policy. Motivated by such empirical analysis, we propose model-based offline ta-RL with regularized policy optimization (MerPO), which learns a meta-model for efficient task structure inference and an informative meta-policy for safe exploration of out-of-distribution state-actions. In particular, we devise a new meta-Regularized model-based Actor-Critic (RAC) method for within-task policy optimization, as a key building block of MerPO, using both conservative policy evaluation and regularized policy improvement; and the intrinsic tradeoff therein is achieved via striking the right balance between two regularizers, one based on the behavior policy and the other on the meta-policy. We theoretically show that the learnt policy offers guaranteed improvement over both the behavior policy and the meta-policy, thus ensuring the performance improvement on new tasks via offline Meta-RL. Our experiments corroborate the superior performance of MerPO over existing offline Meta-RL methods. 
    more » « less
  5. ; ( , IEEE Transactions on Artificial Intelligence)
    The actor-critic RL is widely used in various robotic control tasks. By viewing the actor-critic RL from the perspective of variational inference (VI), the policy network is trained to obtain the approximate posterior of actions given the optimality criteria. However, in practice, the actor-critic RL may yield suboptimal policy estimates due to the amortization gap and insufficient exploration. In this work, inspired by the previous use of Hamiltonian Monte Carlo (HMC) in VI, we propose to integrate the policy network of actor-critic RL with HMC, which is termed as Hamiltonian Policy. As such we propose to evolve actions from the base policy according to HMC, and our proposed method has many benefits. First, HMC can improve the policy distribution to better approximate the posterior and hence reduce the amortization gap. Second, HMC can also guide the exploration more to the regions of action spaces with higher Q values, enhancing the exploration efficiency. Further, instead of directly applying HMC into RL, we propose a new leapfrog operator to simulate the Hamiltonian dynamics. Finally, in safe RL problems, we find that the proposed method can not only improve the achieved return, but also reduce safety constraint violations by discarding potentially unsafe actions. With comprehensive empirical experiments on continuous control baselines, including MuJoCo and PyBullet Roboschool, we show that the proposed approach is a data-efficient and easy-to-implement improvement over previous actor-critic methods. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email