Semi-pessimistic RL tackles distributional shift and data scarcity in offline RL.
problem Distributional shift and scarcity of labeled data in offline RL.
method Proposes a semi-pessimistic RL method that simplifies learning by seeking a lower bound of the reward function.
result Demonstrates clear competitiveness and improved policy learning with vast unlabeled data.
MOReL learns offline RL policies using pessimistic MDPs.
problem Offline RL's data efficiency and velocity.
method Two-step process: learn P-MDP and near-optimal policy in it.
result MOReL is minimax optimal and matches state-of-the-art results.
Paper develops neural network approximation for pessimistic offline RL with theoretical guarantees.
problem Challenges in offline reinforcement learning with deep neural networks and data dependence.
method Establishes estimation error for pessimistic offline RL using neural network approximation with C \mathcal{C} C -mixing data. result Explicit efficiency of deep adversarial offline RL frameworks demonstrated with two converging error components.
Optimistic initialisation is an effective strategy for efficient exploration in reinforcement learning (RL). In the tabular case, all provably efficient model-free algorithms rely on it. However, model-free deep RL algorithms do not use optimistic initialisation despite taking inspiration from these provably efficient …
Papers learn from data to make decisions without interacting, improving on previous methods.
problem Achieving optimal decision-making from offline data with non-linear function approximation.
method Pessimistic Nonlinear Least-Square Value Iteration (PNLSVI) with three innovative components.
result Achieves minimax optimal instance-dependent regret for non-linear function approximation.
Study risk-sensitive RL in offline settings, improving efficiency and accuracy.
problem Efficiently derive near-optimal policies for risk-sensitive RL using offline data.
method Introduced two provably sample-efficient algorithms for risk-sensitive offline RL in linear MDPs.
result First provably efficient risk-sensitive offline RL algorithms.
Pessimistic RL algorithm improves offline RL performance.
problem Insufficient dataset coverage in offline RL.
method Proposes a pessimistic variant of value iteration (PEVI) with a penalty function.
result Establishes upper bound on suboptimality for general MDPs, matching lower bound.
A new algorithm for offline RL with trajectory-wise reward reduces bias and variance errors.
problem Offline RL with trajectory-wise reward incurs large bias and variance errors.
method PARTED algorithm that decomposes trajectory return into proxy rewards and performs pessimistic value iteration.
result PARTED achieves provably efficient suboptimality bounds in general MDPs with trajectory-wise reward.
Pessimistic Q-learning improves sample efficiency in offline reinforcement learning.
problem Insufficient coverage and sample scarcity in offline reinforcement learning datasets.
method Pessimistic Q-learning algorithm for offline reinforcement learning, focusing on variance reduction.
result Near-optimal sample complexity achieved with the proposed algorithm.
This study tackles adversarial corruption in model-based reinforcement learning.
problem Adversarial corruption in model-based reinforcement learning.
method Maximum likelihood estimation (MLE) approach for learning transition model in both online and offline settings.
result Proves a regret of i l d e O ( T + C ) ilde{\mathcal{O}}(\sqrt{T} + C) i l d e O ( T + C ) for CR-OMLE and a suboptimality of O ( C / n ) \mathcal{O}(C/n) O ( C / n ) for CR-PMLE. New methods tackle robust reinforcement learning in sparse, corrupted data.
problem Tackles robust reinforcement learning in sparse, corrupted data.
method Proposes actor-critic methods with sparse robust estimator oracles.
result First non-vacuous guarantees in high-dimensional sparse MDPs with single-policy concentrability coverage.
Offline RL policies should adapt to unknown aspects of the environment.
problem Uncertainty in offline RL datasets leads to suboptimal policies.
method Adaptive policies that consider all transitions seen so far, solving an implicit POMDP.
result Optimal adaptive policies improve offline RL performance.
New offline RL method handles average-reward MDPs with single-policy coverage.
problem Challenges in offline reinforcement learning due to distribution shift and non-uniform coverage.
method Develops an algorithm based on pessimistic discounted value iteration with quantile clipping.
result First fully single-policy sample complexity bound for average-reward offline RL.
Develops methods for dynamic pricing in incomplete data settings.
problem Incomplete historical data makes optimal pricing difficult.
method Nonparametric partial identification framework for offline dynamic pricing.
result Pessimistic and opportunistic policies with regret bounds.
PESCAL uses mediators to learn from confounded offline data.
problem Learning from confounded observational data in reinforcement learning.
method PESCAL uses mediator variables and the pessimistic principle to address confounding bias and distributional shift.
result It is sufficient to learn a lower bound of the mediator distribution function to mitigate distributional shift.
A new algorithm learns policies from batch data in hierarchical RL.
problem Learning policies from fixed batches of data without full exploration.
method Modeling RL as a two-player game with a leader-follower structure, proposing StackelbergLearner.
result StackelbergLearner achieves competitive performance in batch RL and real-world datasets.
Dual active learning improves RLHF by selecting optimal conversations and teachers.
problem Efficiently aligning LLMs with human preferences using RLHF from feedback.
method Offline RL for conversation and teacher selection, dual active reward learning, pessimistic RL.
result The proposed algorithm achieves minimal generalized variance and outperforms state-of-the-arts.
New offline RL method works with limited data and function approximators.
problem Sample efficiency with limited data and weak function approximators.
method Pessimistic algorithm based on version space formed by marginalized importance sampling (MIS), with gap assumption.
result Guarantees sample efficiency for simple algorithm under specific assumptions.
BCPO optimizes offline RL policies by converting uncertainty into conservative bounds.
problem Offline RL's fragility under distribution shifts and model errors.
method Bayesian approach with credible lower bounds and KL regularization.
result BCPO yields an uncertainty-calibrated policy that avoids exploiting model errors.
A new algorithm improves offline reinforcement learning robustness.
problem Finding optimal policies in perturbed environments from offline data.
method Doubly Pessimistic Model-based Policy Optimization (P^2MPO) framework.
result Proves sample efficiency with robust partial coverage data.
New Bellman error estimator improves offline model selection performance.
problem Selecting the best policy from logged data using mean squared Bellman error.
method Developed a more accurate estimator of MSBE and analyzed conditions for successful OMS.
result New estimator achieves impressive offline model selection performance on diverse tasks.
New algorithm tackles confounding in offline RL for partially observable MDPs.
problem Confounding in offline reinforcement learning for partially observable MDPs.
method P3O algorithm using proximal causal inference and pessimistic confidence regions.
result Achieves n − 1 / 2 n^{-1/2} n − 1/2 -suboptimality for offline RL in POMDPs with confounded data. Paper proposes a method to optimize policies for diverse individuals using heterogeneous data.
problem Learning optimal policies for a heterogeneous population from pre-collected data.
method Individualized offline policy optimization framework for heterogeneous MDPs.
result The proposed P4L algorithm achieves a fast rate of average regret.
This paper explains why distributional reinforcement learning is better than vanilla RL using small-loss bounds.
problem Understanding when and why distributional reinforcement learning (DistRL) is superior to vanilla reinforcement learning (RL).
method The paper uses small-loss bounds to explain the benefits of DistRL, proposing algorithms and proving bounds for different RL settings.
result Distributional reinforcement learning (DistRL) outperforms vanilla RL when optimal costs are small, as shown by small-loss bounds.
Efficient offline reinforcement learning with neural networks using differentiable function approximation.
problem Statistical efficiency of offline reinforcement learning with function approximators.
method Pessimistic fitted Q-learning (PFQL) and differentiable function approximation.
result Provably efficient offline reinforcement learning with differentiable function approximation.
New algorithm identifies near-optimal policies in adversarial distributed RL settings.
problem Adversarial agents in distributed RL settings that can collude and report arbitrary data.
method Weighted-Clique algorithm for robust mean estimation from batches, combined with novel distributed algorithms.
result Achieves superior robustness guarantees and near-optimal sample complexities in both offline and online settings.
This paper is concerned with offline reinforcement learning (RL), which learns using pre-collected data without further exploration. Effective offline RL would be able to accommodate distribution shift and limited data coverage. However, prior algorithms or analyses either suffer from suboptimal sample complexities or …
Study proposes a new risk measure for optimal portfolio allocation.
problem Challenges in estimating optimal portfolios based on pessimistic risk.
method Introduces uniform pessimistic risk and computational algorithm.
result Demonstrates the usefulness of the proposed risk and portfolio model with real data analysis.
Pessimistic estimator improves multi-objective policy optimization.
problem Optimizing multi-objective policies from existing data.
method Pessimistic estimator based on inverse propensity scores (IPS).
result Pessimistic estimator outperforms naive IPS estimator in theory and experiments.
Q-Distribution Guided Q-Learning corrects overestimation of uncertain OOD actions in offline RL.
problem Overestimation of Q-values for out-of-distribution actions in offline reinforcement learning.
method QDQ applies a pessimistic adjustment to Q-values in uncertain OOD regions based on a consistency model.
result QDQ improves performance on the D4RL benchmark and achieves significant improvements across many tasks.
Extends linear MDP to handle nonlinear rewards.
problem Restrictive linear MDP assumption limits real-world applicability.
method Proposes Generalized Linear MDP (GLMDP) with GLMs for rewards.
result Develops offline RL algorithms achieving suboptimality guarantees.
New RL algorithms find SNE in Markov games with myopic followers.
problem Finding SNE in Markov games with myopic followers.
method Optimistic and pessimistic variants of least-squares value iteration, incorporating function approximation.
result First provably efficient RL algorithms for SNEs in general-sum Markov games with myopic followers.
FedLCB-Q learns optimal policies from federated offline data with linear speedup.
problem Learning optimal policies from offline data with federated learning.
method Federated offline RL algorithm tailored for Q-learning, using local Q-function updates and central aggregation.
result Achieves linear speedup in sample complexity with collaboration among agents.
Paper tackles constrained bandit problems with a new learning framework.
problem Optimizing a black-box reward function subject to a black-box constraint function over a continuous space.
method Rectified Pessimistic-Optimistic Learning (RPOL) framework, incorporating optimistic and pessimistic GP bandit learning.
result RPOL achieves sublinear regret and minimal cumulative constraint violation.
Paper tackles RLHF with DCPPO method, proving near-optimal suboptimality.
problem Challenges in offline RLHF with limited human feedback and bounded rationality.
method DCPPO method involving three stages: MLE, reward function recovery, and pessimistic value iteration.
result DCPPO's suboptimality almost matches classical pessimistic offline RL in terms of distribution shift and dimension.
New RL algorithms learn QSE from strategic feedbacks with sample efficiency.
problem Learning QSE in Markov games with strategic feedbacks.
method Proposes sample-efficient algorithms for online and offline settings, combining quantal response model learning and RL.
result Achieves sublinear regret bounds and quantifies model uncertainty.
New algorithm reduces offline RL sample complexity for MDPs.
problem Learning optimal policies from offline data in unknown MDPs.
method Adaptive Pessimistic Value Iteration (APVI) algorithm.
result Suboptimality bound nearly matches theoretical limits.
New algorithm learns optimal policies in strategic MDPs with private types.
problem Optimal policy learning in strategic MDPs with private types and information asymmetry.
method PLAN algorithm using instrumental variable regression and pessimism principle.
result PLAN achieves near-optimal policy with 1 / K 1 / \sqrt{K} 1/ K optimality. Pessimistic Minimax Value Iteration finds efficient NE policies from offline data.
problem Finding an approximate Nash equilibrium in offline Markov games with non-uniform coverage.
method Pessimistic Minimax Value Iteration (PMVI) constructs pessimistic value function estimates and solves NEs.
result Established a nearly minimax optimal result for offline Markov games with function approximation.
Survey of reinforcement learning guarantees with data constraints.
problem Guaranteeing near-optimal policies with limited data in reinforcement learning.
method Coverage-Structure-Objective (CSO) framework to decompose sample complexity results.
result Progress on PAC guarantees for reinforcement learning, covering various models and settings.
POLAR optimizes treatment strategies in dynamic settings with statistical guarantees.
problem Optimizing sequential decisions in dynamic treatment regimes with robustness and statistical guarantees.
method Pessimistic model-based approach estimating transition dynamics and incorporating uncertainty penalties.
result Offers statistical and computational guarantees, including finite-sample bounds on policy suboptimality.
This paper develops a model of reference-dependent assessment of subjective beliefs in which loss-averse people optimally choose the expectation as the reference point to balance the current felicity from the optimistic anticipation and the future disappointment from the realisation. The choice of over-optimism or over…
This work improves RLHF performance guarantees using greedy sampling.
problem Learning the KL-regularized target with preference feedback.
method General preference model, greedy sampling.
result Order-wise improvements over existing RLHF performance guarantees.
Proposes DRRO to mitigate over-optimization in RLHF from human feedback.
problem Over-optimization due to reward misspecification in RLHF.
method Wasserstein distributionally robust regret optimization (DRRO).
result DRRO mitigates over-optimization more effectively than existing baselines.
PASTA optimizes assortment selection using pessimism principle.
problem Optimizing assortment selection with limited data coverage.
method Pessimistic Assortment Optimization (PASTA) based on the principle of pessimism.
result PASTA correctly identifies optimal assortment with minimal data coverage.
Trajectory-level supervision allows efficient offline reinforcement learning.
problem Offline reinforcement learning
method Developing a statistical theory for offline policy optimization from trajectory-level labels
result Proving a high-probability guarantee of order O ~ ( H 2 C s a ( π ⋆ ) / n ) \widetilde O(H^2\sqrt{C_{sa}(\pi^\star)/n}) O ( H 2 C s a ( π ⋆ ) / n ) Study optimal product assortment using historical data, proving item coverage suffices.
problem Offline assortment optimization under MNL model with limited historical data.
method Pessimistic Rank-Breaking (PRB) algorithm combining rank-breaking and pessimistic estimation.
result Optimal item coverage is both sufficient and necessary for efficient offline learning.
This paper investigates to what extent one can improve reinforcement learning algorithms. Our study is split in three parts. First, our analysis shows that the classical asymptotic convergence rate O ( 1 / N ) O(1/\sqrt{N}) O ( 1/ N ) is pessimistic and can be replaced by O ( ( log ( N ) / N ) β ) O((\log(N)/N)^β) O (( log ( N ) / N ) β ) with 1 2 ≤ β ≤ 1 \frac{1}{2}\leq β\leq 1 2 1 ≤ β ≤ 1 and N N N the number…