Continuous Doubly Constrained Batch Reinforcement Learning

Fakoor, Rasool; Mueller, Jonas; Asadi, Kavosh; Chaudhari, Pratik; Smola, Alexander J.

doi:10.48550/arxiv.2102.09225

Cited by 3 publications

(4 citation statements)

References 24 publications

(37 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In practice, we can use the KL divergence to replace the total variation distance between policies, based on Pinsker's Inequality: π 1 −π 2 ≤ 2D KL (π 1 ||π 2 ). Moreover, since the behavior policy π β,n is typically unknown, we can use the reverse KL-divergence between π n and π β,n to circumvent the estimation of π β,n , following the same line as in (Fakoor et al, 2021):…”

Section: Safe Policy Improvement With Meta-regularizationmentioning

confidence: 99%

Model-Based Offline Meta-Reinforcement Learning with Regularization

Shen¹,

Wan²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

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 explore model-based offline Meta-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 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. Experiments corroborate the superior performance of MerPO over existing offline Meta-RL methods.

show abstract

Section: Safe Policy Improvement With Meta-regularizationmentioning

confidence: 99%

Model-Based Offline Meta-Reinforcement Learning with Regularization

Shen¹,

Wan²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Insufficient coverage of the dataset due to the lack of online exploration is known as the main challenge in offline RL . To deal with this problem, a number of methods have been recently proposed from both model-free (Wu et al, 2019;Touati et al, 2020;Liu et al, 2020;Rezaeifar et al, 2021;Fujimoto et al, 2019;Fakoor et al, 2021) and model-based perspectives (Yu et al, 2020;Kidambi et al, 2020;Matsushima et al, 2020;Yin et al, 2021). More or less, their methods rely on the idea of pessimism and its variants in the sense that the learned policy can avoid uncertain regions not covered by offline data.…”

Section: Related Workmentioning

confidence: 99%

Pessimistic Model-based Offline Reinforcement Learning under Partial Coverage

Uehara¹,

Sun²

2021

Preprint

View full text Add to dashboard Cite

We study model-based offline Reinforcement Learning with general function approximation. We present an algorithm named Constrained Pessimistic Policy Optimization (CPPO) which leverages a general function class and uses a constraint to encode pessimism. Under the assumption that the ground truth model belongs to our function class, CPPO can learn with the offline data only providing partial coverage, i.e., it can learn a policy that completes against any policy that is covered by the offline data, in polynomial sample complexity with respect to the statistical complexity of the function class. We then demonstrate that this algorithmic framework can be applied to many specialized Markov Decision Processes where the additional structural assumptions can further refine the concept of partial coverage. One notable example is low-rank MDP with representation learning where the partial coverage is defined using the concept of relative condition number measured by the underlying unknown ground truth feature representation. Finally, we introduce and study the Bayesian setting in offline RL.The key benefit of Bayesian offline RL is that algorithmically, we do not need to explicitly construct pessimism or reward penalty which could be hard beyond models with linear structures. We present a posterior sampling based incremental policy optimization algorithm (PS-PO) which proceeds by iteratively sampling a model from the posterior distribution and performing one step incremental policy optimization inside the sampled model. Theoretically, in expectation with respect to the prior distribution, PS-PO can learn a near optimal policy under partial coverage with polynomial sample complexity.

show abstract

“…Offline RL In offline RL, algorithms such as FQI (Ernst et al, 2005) have finite-sample error guarantees under the global coverage Antos et al, 2008). Recently, many algorithms to tackle this problem have been proposed from both model-free (Wu et al, 2019;Touati et al, 2020;Liu et al, 2020;Fujimoto et al, 2019;Fakoor et al, 2021;Kumar et al, 2020) and model-based perspectives Kidambi et al, 2020;Matsushima et al, 2020) with some pessimism ideas. The idea of pessimism features in offline RL with an eye to penalize the learner from visiting unknown regions of the state-action space (Rashidinejad et al, 2021;Yin et al, 2021;Buckman et al, 2020).…”

Section: Related Workmentioning

confidence: 99%

“…The first other way is imposing constraints on the policy class or Q-function class so that estimated policies are not too much far away from behavior policies. For example, we can use KL divegences, MMD distance, Wasserstein distance to measure the distance from behavior policies (Wu et al, 2019;Fakoor et al, 2021;Matsushima et al, 2020;Touati et al, 2020;Fujimoto et al, 2019) and add D(π, π b ) as penalty terms, where π b is a behavior policy. Another way is explicitly estimating the lower bound of q-functions (Kumar et al, 2020;.…”

Section: Now We Only Need To Focus On Analyzingmentioning

confidence: 99%

Mitigating Covariate Shift in Imitation Learning via Offline Data Without Great Coverage

Chang

Uehara

Sreenivas

et al. 2021

Preprint

View full text Add to dashboard Cite

This paper studies offline Imitation Learning (IL) where an agent learns to imitate an expert demonstrator without additional online environment interactions. Instead, the learner is presented with a static offline dataset of state-action-next state transition triples from a potentially less proficient behavior policy. We introduce Model-based IL from Offline data (MILO): an algorithmic framework that utilizes the static dataset to solve the offline IL problem efficiently both in theory and in practice. In theory, even if the behavior policy is highly sub-optimal compared to the expert, we show that as long as the data from the behavior policy provides sufficient coverage on the expert state-action traces (and with no necessity for a global coverage over the entire state-action space), MILO can provably combat the covariate shift issue in IL. Complementing our theory results, we also demonstrate that a practical implementation of our approach mitigates covariate shift on benchmark MuJoCo continuous control tasks. We demonstrate that with behavior policies whose performances are less than half of that of the expert, MILO still successfully imitates with an extremely low number of expert state-action pairs while traditional offline IL method such as behavior cloning (BC) fails completely. Source code is provided at https://github.com/jdchang1/milo. * Equal contribution † Work done outside of Amazon 3 in this work, we use cost instead of reward, thus we call A π disadvantage function.

show abstract

Continuous Doubly Constrained Batch Reinforcement Learning

Cited by 3 publications

References 24 publications

Model-Based Offline Meta-Reinforcement Learning with Regularization

Model-Based Offline Meta-Reinforcement Learning with Regularization

Pessimistic Model-based Offline Reinforcement Learning under Partial Coverage

Mitigating Covariate Shift in Imitation Learning via Offline Data Without Great Coverage

Contact Info

Product

Resources

About