BRPO optimizes batch RL policies to better exploit state-action differences.
problem Batch RL's conservatism limits exploitation of state-action differences.
method Proposes residual policies and derives BRPO to maximize policy performance.
result BRPO achieves state-of-the-art performance in various tasks.
Policy gradient methods with aggregated states can achieve better performance than approximate policy iteration.
problem Approximation errors in policy and value function approximations.
method State-aggregated representations and policy gradient methods.
result Policy gradient methods can achieve a per-period regret bounded by ε, while approximate policy iteration and value iteration have a higher regret.
Off-policy deep reinforcement learning (RL) algorithms are incapable of learning solely from batch offline data without online interactions with the environment, due to the phenomenon known as \textit{extrapolation error}. This is often due to past data available in the replay buffer that may be quite different from th…
New RL method learns value function for many policies using few key states.
problem Evaluate and improve policies in continuous control problems.
method Combines actor-critic architecture and policy embedding to learn a single value function for many policies.
result Value function minimizes prediction error by learning a small set of 'probing states' and their impact on policies' returns.
We consider off-policy policy evaluation when the trajectory data are generated by multiple behavior policies. Recent work has shown the key role played by the state or state-action stationary distribution corrections in the infinite horizon context for off-policy policy evaluation. We propose estimated mixture policy …
We study the problem of off-policy policy optimization in Markov decision processes, and develop a novel off-policy policy gradient method. Prior off-policy policy gradient approaches have generally ignored the mismatch between the distribution of states visited under the behavior policy used to collect data, and what …
Paper proposes a policy-search algorithm to learn entropy-maximizing exploration policies in reward-free environments.
problem Reward-free learning in high-dimensional, continuous-control domains.
method Maximum Entropy POLicy optimization (MEPOL) algorithm that maximizes a non-parametric state entropy estimate.
result MEPOL learns a maximum-entropy exploration policy that facilitates learning various reward-based tasks.
PBVFs generalize across policies using learned value functions.
problem RL algorithms forget information about old policies when updating value functions to track the learned policy.
method Introduce Parameter-Based Value Functions (PBVFs) that include policy parameters in their inputs, enabling them to generalize across different policies.
result PBVFs enable zero-shot learning of new policies that outperform any policy seen during training.
Paper introduces a new value function for state transitions and optimal policy learning.
problem Learning optimal policies from state transitions and actions.
method Develops a forward dynamics model to maximize a novel value function Q(s,s′). result Demonstrates benefits in value function transfer, redundant action spaces, and off-policy learning.
Opioid overdose rates have reached an epidemic level and state-level policy innovations have followed suit in an effort to prevent overdose deaths. State-level drug law is a set of policies that may reinforce or undermine each other, and analysts have a limited set of tools for handling the policy collinearity using st…
RPO uses past and future state-action info for better policy optimization.
problem Sample inefficiency in on-policy reinforcement learning methods.
method Reflective Policy Optimization (RPO) integrates past and future state-action info for policy improvement.
result RPO improves policy performance and contracts the solution space, leading to faster convergence.
The policy gradient theorem is defined based on an objective with respect to the initial distribution over states. In the discounted case, this results in policies that are optimal for one distribution over initial states, but may not be uniformly optimal for others, no matter where the agent starts from. Furthermore, …
Entropy regularization is used to get improved optimization performance in reinforcement learning tasks. A common form of regularization is to maximize policy entropy to avoid premature convergence and lead to more stochastic policies for exploration through action space. However, this does not ensure exploration in th…
This paper simplifies OPE in large state spaces using state abstractions.
problem Accurately evaluating policies offline in large state spaces.
method Developed a backward-model-irrelevance condition and an iterative state abstraction procedure.
result Deeply-abstracted states substantially simplify OPE sample complexity.
This paper improves reinforcement learning policies in a scalable way.
problem Ensuring monotonic policy improvement in entropy-regularized RL.
method Derives an entropy-aware lower bound and proposes a novel RL algorithm.
result Demonstrates effectiveness in continuous-state tasks using a linear function approximator.
Optimizes state monitoring in Markovian systems with cost constraints.
problem Balancing state queries with prediction costs in Markovian systems.
method Greedy policy and SGD-based learning variant for optimal predict-query tradeoff.
result Greedy policy is suboptimal but performs close to optimal under certain conditions.
We introduce Recurrent Predictive State Policy (RPSP) networks, a recurrent architecture that brings insights from predictive state representations to reinforcement learning in partially observable environments. Predictive state policy networks consist of a recursive filter, which keeps track of a belief about the stat…
Policy analysts wish to visualize a range of policies for large simulator-defined Markov Decision Processes (MDPs). One visualization approach is to invoke the simulator to generate on-policy trajectories and then visualize those trajectories. When the simulator is expensive, this is not practical, and some method is r…
New method optimizes policies in non-stationary environments.
problem Optimizing policies in non-stationary, context-dependent environments.
method Two-phase approach: offline learning and online adaptation.
result Our method outperforms existing approaches in both synthetic and real-world datasets.
Simplifies complex RL policies by ranking important decisions.
problem Complexity in RL policies makes them hard to analyze and interpret.
method Statistical fault localisation to rank states and prune unimportant decisions.
result Pruned policies can perform similarly to original policies, improving interpretability.
OSIRIS reduces variance in off-policy evaluation by omitting irrelevant states.
problem High variance in importance sampling-based OPE estimators.
method OSIRIS reduces variance by omitting likelihood ratios associated with states irrelevant to return.
result OSIRIS is unbiased and has lower variance than ordinary importance sampling.
Sample inefficiency is a long-lasting problem in reinforcement learning (RL). The state-of-the-art estimates the optimal action values while it usually involves an extensive search over the state-action space and unstable optimization. Towards the sample-efficient RL, we propose ranking policy gradient (RPG), a policy …
Improved SAC with AWMP for better control tasks.
problem Discontinuous and non-smooth optimal policies in reinforcement learning.
method Advantage Weighted Mixture Policy (AWMP) for SAC, learning state-specific weights.
result SAC with AWMP outperforms SAC in four control tasks.
New approach to understand recurrent policies as FSMs without minimization.
problem Minimization of FSMs obscures the semantics of policy decisions.
method Start with unminimized FSM, apply interpretable reductions, use attention tool.
result Reveals insights into policy decisions not previously noticed.
Paper proposes interpretable RL policies from a mixture of experts.
problem Making RL policies transparent and understandable in real-world applications.
method Policy iteration scheme with interpretable experts and prototypical states.
result Proposed algorithm learns policies comparable to neural networks but more interpretable.
MAPS and MAPS-SE learn from multiple suboptimal experts to improve policies efficiently.
problem Learning from multiple suboptimal experts in reinforcement learning.
method Active policy improvement from multiple black-box oracles.
result MAPS and MAPS-SE achieve sample efficiency and state-wise imitation learning.
Stabilizes policy optimization with off-policy data using divergence augmentation.
problem Premature convergence and instability in policy optimization with off-policy data.
method Incorporates Bregman divergence between behavior and current policies to ensure safe policy updates.
result Empirically shows better performance in data-scarce scenarios compared to other algorithms.
We develop a normative framework for hierarchical model-based policy optimization based on applying second-order methods in the space of all possible state-action paths. The resulting natural path gradient performs policy updates in a manner which is sensitive to the long-range correlational structure of the induced st…
New method improves reinforcement learning generalization.
problem Few environments lead to poor generalization in reinforcement learning.
method Integrates sequential structure into representation learning, using a policy similarity metric (PSM) and contrastive embeddings (PSEs).
result PSEs improve generalization across various benchmarks.
SCPO learns robust policies without modeling disturbance, improving real-world task performance.
problem Poor performance of reinforcement learning in real-world tasks due to disturbance in transition dynamics.
method State-conservative policy optimization (SCPO) that reduces disturbance to state space and approximates it with a gradient-based regularizer.
result SCPO learns robust policies without prior knowledge of disturbance or simulators, improving performance in robot control tasks.
This paper investigates the resilience and robustness of Deep Reinforcement Learning (DRL) policies to adversarial perturbations in the state space. We first present an approach for the disentanglement of vulnerabilities caused by representation learning of DRL agents from those that stem from the sensitivity of the DR…
This paper introduces a new method to evaluate multiple policies simultaneously.
problem Estimating the value of many policies for a single set of states.
method Developed a scalable, differentiable fingerprinting mechanism to represent complex policies.
result The method can produce policies that outperform those that generated the training data, in zero-shot manner.
New RL approach uses future state and action visitation measures for better exploration.
problem Improving exploration in reinforcement learning.
method Intrinsic reward based on future state and action visitation measures, using contraction operators.
result Policies achieve good state-action space coverage and high performance.
We decode latent states in Block MDPs and learn near-optimal policies.
problem Model estimation and reward-free learning in Block MDPs.
method Information-theoretical lower bound and efficient model estimation algorithm.
result Our algorithm approaches the information-theoretical limit for latent state decoding and converges to optimal policies.
New method recovers diverse policies from expert data using state-action pair weighting.
problem Recovering diverse policies from expert trajectories.
method Pointwise mutual information weighted behavioral cloning.
result Effective in focusing on state-action pairs most representative of the style.
Study agnostic RL in large state spaces with weak function approximation.
problem Statistical intractability of agnostic policy learning in various environments.
method Investigates agnostic policy learning with different forms of environment access.
result Agnostic policy learning remains statistically intractable with certain forms of environment access.
This paper optimizes MDP policies for efficient state aggregation.
problem Optimizing policies in aggregated Markov chains while preserving optimal performance.
method Homomorphic mappings to establish optimal policy equivalence and derive performance bounds.
result Developed HPG and EBHPG methods for efficient aggregation and policy optimization.
Recent studies have revealed that neural network-based policies can be easily fooled by adversarial examples. However, while most prior works analyze the effects of perturbing every pixel of every frame assuming white-box policy access, in this paper we take a more restrictive view towards adversary generation - with t…
Several approximate policy iteration schemes without value functions, which focus on policy representation using classifiers and address policy learning as a supervised learning problem, have been proposed recently. Finding good policies with such methods requires not only an appropriate classifier, but also reliable e…
A nonparametric approach for policy learning for POMDPs is proposed. The approach represents distributions over the states, observations, and actions as embeddings in feature spaces, which are reproducing kernel Hilbert spaces. Distributions over states given the observations are obtained by applying the kernel Bayes' …
New framework improves sample efficiency and robustness in RL with smooth policies.
problem Sample inefficiency and lack of robustness in deep reinforcement learning.
method SR^2L framework, smoothness-inducing regularization.
result Improved sample efficiency and robustness in both on-policy and off-policy RL algorithms.
POLAR learns efficient data acquisition policies using pretrained belief representations.
problem Challenges in learning effective policies for adaptive data acquisition.
method POLAR decouples representation learning from policy learning by leveraging pretrained predictive foundation models as belief-state encoders.
result POLAR outperforms state-of-the-art methods across diverse tasks while requiring fewer training samples.
New method for natural policy gradients converges linearly.
problem Improving natural policy gradient methods for better convergence.
method Fisher-Rao gradient flow applied to state-action distributions.
result Linear convergence rate with geometry-dependent factor.
New algorithm stabilizes RL policy learning through divergence regularization.
problem Stabilize policy learning and improve performance in RL.
method Proximity term constraining discounted state-action visitation distributions to be close to each other.
result Proposed algorithm improves stability and final performance in RL tasks.
RANDPOL uses randomized networks for efficient reinforcement learning in continuous state and action MDPs.
problem Efficient reinforcement learning in environments with continuous state and action spaces.
method RANDPOL uses randomized function approximation to represent policy and value functions, providing finite time guarantees and improved numerical performance.
result RANDPOL achieves better numerical performance and provides finite time guarantees compared to deep neural network based algorithms.
Reinforcement learning agents that operate in diverse and complex environments can benefit from the structured decomposition of their behavior. Often, this is addressed in the context of hierarchical reinforcement learning, where the aim is to decompose a policy into lower-level primitives or options, and a higher-leve…
The paper explores learning good policies from past data in large state spaces.
problem Learning good policies from historical data in large state spaces.
method Introduces expressivity assumptions and data coverage for function approximation and algorithmic design.
result A variety of algorithms and their guarantees are presented based on assumptions and desired complexity.
New RL method learns from state transitions without actions.
problem Offline RL with missing action labels.
method State policy discretisation and decQN algorithm.
result Improves convergence speed and performance in online RL.