Proposes a new Bayesian learning method for optimal treatment regimes.
problem Sub-optimal policies in offline data due to lack of exploration.
method Integrates pessimism principle with Thompson sampling and Bayesian machine learning.
result Derives a credible set that uniformly lower bounds the optimal Q-function.
The paper introduces Bellman-consistent pessimism to improve offline reinforcement learning without overly pessimistic bias.
problem Offline reinforcement learning's challenge of discovering good policies without exhaustive exploration.
method Introduces Bellman-consistent pessimism for function approximation, improving sample complexity and adaptability.
result Improves sample complexity by O(d) in the action space finite case, and automatically adapts to bias-variance tradeoff. USAC balances pessimism and optimism in actor-critic training for better exploration and performance.
problem Excessive pessimism limits exploration, while excessive optimism leads to high-risk behaviors.
method Utility Soft Actor-Critic (USAC) dynamically adapts exploration based on critic uncertainty.
result USAC consistently outperforms state-of-the-art algorithms in continuous control tasks.
Unified framework for analyzing pessimism in off-policy learning with regularized importance sampling.
problem High variance in importance weighting for off-policy learning.
method Unified PAC-Bayesian study of pessimism with regularized importance sampling.
result Derivation of a tractable PAC-Bayesian generalization bound for common importance weight regularizations.
The paper improves Q-learning by incorporating pessimism for better sample efficiency.
problem Improving sample efficiency in asynchronous Q-learning with non-i.i.d. data.
method Developed an algorithmic framework that incorporates the principle of pessimism into asynchronous Q-learning, penalizing infrequently-visited state-action pairs based on suitable lower confidence bounds (LCBs).
result Achieved near-optimal sample complexity, providing theoretical support for the use of pessimism in non-i.i.d. data.
New method optimizes offline linear bandits using different confidence sets.
problem Optimizing offline learning for linear contextual bandits.
method Introduces a family of pessimistic learning rules based on ℓp confidence sets. result The π^∞ rule achieves minimax performance and strictly dominates other predictors. 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 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.
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.
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.
New model explains price dynamics of Bitcoin with psychological factors.
problem Understanding price variations in cryptocurrency markets with psychological factors.
method Extended agent-based model with heterogeneous psychological parameters.
result Model shows diverse dynamics based on psychological correlation.
Bayesian approach to portfolio selection reduces pessimism in frequent trading.
problem Tackling the challenge of estimating drift in Merton's portfolio selection model.
method Bayesian distributionally robust control with nonlinear Wasserstein projections.
result Reduced pessimism and improved performance in frequent rebalancing compared to existing methods.
Study examines how pandemic anxiety affects financial market trust.
problem Anxiety during pandemic and trust in financial markets.
method Used Google search volume and stock market data to create mood indicators.
result Different clusters of countries and markets in terms of pessimism and optimism emerged.
New assumptions and algorithm solve offline two-player zero-sum Markov games.
problem Solving offline two-player zero-sum Markov games under insufficient assumptions.
method Proposed unilateral concentration assumption and pessimism-type algorithm.
result Algorithm efficiently learns Nash equilibrium under unilateral concentration.
New method reduces over-pessimism in Bayesian control under parameter uncertainty.
problem Over-pessimism in Bayesian control due to misspecified priors.
method Distributionally robust Bayesian control (DRBC) with strong duality and optimization.
result Validated algorithm on synthetic and real data, reducing over-pessimism.
Latent variable models improve RL by facilitating efficient learning and exploration.
problem Improving sample efficiency in reinforcement learning.
method Representation view of latent variable models for state-action value functions, incorporating kernel embeddings and UCB exploration.
result Established sample complexity of the proposed approach in online and offline settings, demonstrated superior performance in benchmarks.
Distributionally Robust Supervised Learning (DRSL) is necessary for building reliable machine learning systems. When machine learning is deployed in the real world, its performance can be significantly degraded because test data may follow a different distribution from training data. DRSL with f-divergences explicitly …
LoCo-RLHF models diverse human feedback with contextual information.
problem Heterogeneous human feedback from diverse contexts and preferences.
method Low-rank contextual preference model, PRS policy.
result LoCo-RLHF achieves tighter sub-optimality gap than existing methods.
New method tackles MDPs by learning normalized representations efficiently.
problem Curse of dimensionality in MDPs.
method Contrastive representation learning for linear MDPs.
result First practical method with strong theoretical guarantees and empirical performance.
RH-UCRL combines pessimism and optimism for robust RL.
problem Ensuring reliable performance in real-world RL tasks with worst-case scenarios.
method RH-UCRL is a model-based RL algorithm that optimizes between an agent and an adversary, distinguishing between epistemic and aleatoric uncertainty.
result RH-UCRL achieves near-optimal sample complexity guarantees and outperforms other robust RL algorithms in adversarial environments.
This work improves policy evaluation and selection using logarithmic smoothing for pessimistic off-policy estimation.
problem Offline evaluation and selection of policies from past data.
method Develops novel concentration bounds and a logarithmically smoothed estimator (LS) for improved policy selection and learning.
result The logarithmically smoothed estimator (LS) provides tighter bounds and better policy selection and learning.
Paper tackles efficient IRL in offline settings with polynomial samples and runtime.
problem Efficiently learning reward functions from expert demonstrations in offline settings.
method Adapting the pessimism principle for offline RL, achieving strong guarantees.
result Achieved efficient IRL in offline and online settings with polynomial samples and runtime.
New algorithm for offline RL with linear approx in MDPs and MGs, nearly optimal.
problem Offline RL with linear function approximation in MDPs and MGs.
method Pessimism-based algorithm with uncertainty decomposition via reference function.
result Nearly minimax optimal performance in offline RL for MDPs and MGs.
Algorithm learns optimal dynamic mechanisms from data.
problem Designing optimal mechanisms for dynamic settings with unknown reward functions.
method Offline reinforcement learning with pessimism principle.
result Learned mechanisms are efficient, individually rational, and truthful.
Unified approach to RLHF tackles uncertainty in reward function.
problem Uncertainty in reward function learned from human feedback.
method Value-incentivized preference optimization (VPO) that regularizes the reward function with value function.
result Theoretical and practical guarantees for both online and offline RLHF settings.
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.
New method boosts skill learning by encouraging optimistic exploration.
problem Intrinsic reward for exploration is inherently pessimistic.
method Derive an information gain auxiliary objective involving an ensemble of discriminators and rewarding policy disagreement.
result Improves skill learning in grid worlds and Atari games.
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…
New algorithm CROP achieves asymptotic optimality with bounded regret.
problem Optimistic algorithms fail to achieve asymptotic instance-dependent regret optimality.
method CRush Optimism with Pessimism (CROP) algorithm that eliminates optimistic hypotheses.
result CROP achieves constant-factor asymptotic optimality and bounded regret.
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.
In the presence of model risk, it is well-established to replace classical expected values by worst-case expectations over all models within a fixed radius from a given reference model. This is the "robustness" approach. We show that previous methods for measuring this radius, e.g. relative entropy or polynomial diverg…
New algorithm improves inference-time alignment without reward hacking.
problem Improving quality of responses from language models with limited compute.
method Inference-time alignment, focusing on extttInferenceTimePessimism algorithm. result Optimal performance and scaling-monotonicity of extttInferenceTimePessimism. 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.
Study human-machine interaction with private info using offline RL.
problem Confounding bias and distributional mismatch in offline RL for human-guided interaction.
method Developed a novel identification result and OPE method to address confounding bias, and used pessimism to tackle distributional mismatch.
result Policy pair converges to optimal one at satisfactory rate under mild assumptions.
A new offline RL framework unifies imitation learning and vanilla offline RL.
problem Learning from expert datasets without active data collection.
method A new offline RL framework that interpolates between imitation learning and vanilla offline RL, using a weak concentrability coefficient and a lower confidence bound algorithm.
result LCB algorithm achieves a faster rate of 1/N for nearly-expert datasets, and is adaptively optimal for the entire data composition range. Novel framework proves fast RL convergence in continuous spaces.
problem Analyzing stability in continuous state-action RL.
method Introduces a novel framework to analyze stability properties of RL.
result Highlights two key stability properties and demonstrates their satisfaction in RL.
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.
Paper tackles robust offline RL with heavy-tailed rewards.
problem Real-world applications often encounter heavy-tailed rewards, challenging offline RL.
method Proposes ROAM and ROOM algorithms using median-of-means method for robust off-policy evaluation and OPO.
result Demonstrates superior performance on heavy-tailed reward datasets compared to existing methods.
We provide an economic interpretation of the practice consisting in incorporating risk measures as constraints in a classic expected return maximization problem. For what we call the infimum of expectations class of risk measures, we show that if the decision maker (DM) maximizes the expectation of a random return unde…
Proposes a method to improve pWCET estimation for heavy-tailed distributions.
problem Improving pWCET estimation for heavy-tailed distributions in real-time systems.
method Incorporates saturating functions into Chebyshev's inequality to mitigate the influence of large outliers.
result Achieves safe and tighter bounds for heavy-tailed distributions.
A new algorithm CAP learns optimal policies from observational data with confounding bias and missing observations.
problem Offline contextual bandit with confounding bias and missing observations.
method CAP policy learning, forming reward function as solution of integral equation system, building confidence set, and greedily taking action with pessimism.
result Developed an upper bound to the suboptimality of CAP for the offline contextual bandit problem.
The policy gradient theorem describes the gradient of the expected discounted return with respect to an agent's policy parameters. However, most policy gradient methods drop the discount factor from the state distribution and therefore do not optimize the discounted objective. What do they optimize instead? This has be…
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.
New algorithm reduces best-in-class regret in contextual bandits.
problem Compete with the best policy in a class without model restrictions.
method Proposes an algorithm that updates policies by minimizing a pessimistic objective, including a clipped inverse-propensity estimate and variance penalty.
result Achieves fast best-in-class regret rates, including polylogarithmic rates in the parametric case.
PPO algorithm converges to global optimality in multi-agent reinforcement learning.
problem Designing statistical guarantees for policy optimization methods in multi-agent reinforcement learning.
method Leveraging a multi-agent performance difference lemma, a localized action value function is used as a descent direction for each local policy, leading to a multi-agent PPO algorithm.
result The multi-agent PPO algorithm converges to the globally optimal policy at a sublinear rate under standard regularity conditions.
Following our previous investigation of the USA Standard and Poor index anti-bubble that started in August 2000, we analyze thirty eight world stock market indices and identify 21 anti-bubble. An ``anti-bubble'' is defined as a self-fulfilling decreasing price created by positive price-to-price feedbacks feeding overal…
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.
AE-LSVI identifies near-optimal policies in complex systems with minimal data.
problem Identifying near-optimal policies in complex, costly data acquisition systems.
method Combines optimism and pessimism for active exploration in a generative model setting.
result Proves near-optimal policy identification over entire state spaces with polynomial sample complexity.