arXiv research

Paper generalizes reward distribution in multi-armed bandits with temporally-partitioned rewards.

Novel framework for risk-sensitive reinforcement learning using martingale decomposition.

The paper studies reward concentration in MDPs, covering asymptotic and non-asymptotic settings.

Curiosity-Critic improves world model training by focusing on cumulative prediction error.

This work extends reinforcement learning to handle non-cumulative objectives.

The paper tackles non-cumulative objectives in reinforcement learning and proposes modifications to existing algorithms.

The stochastic multi-armed bandit (MAB) problem is a common model for sequential decision problems. In the standard setup, a decision maker has to choose at every instant between several competing arms, each of them provides a scalar random variable, referred to as a "reward." Nearly all research on this topic consider…

EgalMAB solves fair resource allocation in stochastic bandits.

The paper addresses human-like decision-making in multi-agent systems using bounded risk-sensitive Markov Games.

Experience replay enables reinforcement learning agents to memorize and reuse past experiences, just as humans replay memories for the situation at hand. Contemporary off-policy algorithms either replay past experiences uniformly or utilize a rule-based replay strategy, which may be sub-optimal. In this work, we consid…

New RL formulation for maximizing maximum reward in molecule generation.

Paper tackles constrained bandit problems with a new learning framework.

Study tackles infinitely many-armed bandits with rotting rewards, achieving tight regret bounds.

Bayesian algorithms minimize cumulative regret in decentralized multi-agent bandits.

Paper proposes a framework for reliable off-policy evaluation in reinforcement learning.

Optimal strategy proposed for maximizing cumulative reward in continuum-armed bandits.

In this paper, we study reinforcement learning (RL) algorithms to solve real-world decision problems with the objective of maximizing the long-term reward as well as satisfying cumulative constraints. We propose a novel first-order policy optimization method, Interior-point Policy Optimization (IPO), which augments the…

A new method optimizes in nonstationary environments with many arms efficiently.

LinConTS improves regret and constraint violations in probabilistic linearly constrained bandits.

Bayesian optimization improved for biased data.

A novel approach for safe offline RL using latent safety constraints.

UCRL-CMDP algorithm optimizes RL with constraints on average costs.

Algorithm improves multi-armed bandit performance by transferring reward samples.

We consider a novel stochastic multi-armed bandit setting, where playing an arm makes it unavailable for a fixed number of time slots thereafter. This models situations where reusing an arm too often is undesirable (e.g. making the same product recommendation repeatedly) or infeasible (e.g. compute job scheduling on ma…

GACBO optimizes unknown causal graphs with interventions.

Novel approach to contextual bandits using self-supervised learning.

We propose an online algorithm for cumulative regret minimization in a stochastic multi-armed bandit. The algorithm adds $O(t)$ i.i.d. pseudo-rewards to its history in round $t$ and then pulls the arm with the highest average reward in its perturbed history. Therefore, we call it perturbed-history exploration (PHE). Th…

New method reduces dynamic regret for non-stationary bandits.

Curriculum learning has been successfully used in reinforcement learning to accelerate the learning process, through knowledge transfer between tasks of increasing complexity. Critical tasks, in which suboptimal exploratory actions must be minimized, can benefit from curriculum learning, and its ability to shape explor…

New TS algorithms improve performance in non-stationary multi-armed bandit problems.

A new bandit problem where experiments can be interrupted if results are not promising.

Imitation Learning describes the problem of recovering an expert policy from demonstrations. While inverse reinforcement learning approaches are known to be very sample-efficient in terms of expert demonstrations, they usually require problem-dependent reward functions or a (task-)specific reward-function regularizatio…

Neural Thompson Sampling uses deep neural networks for contextual bandit problems.

This work analyzes the value of future reward information in RL.

New algorithms for efficient causal interventions with budget constraints and without constraints.

New method tackles safe reinforcement learning from offline data.

LNUCB-TA improves MAB performance by dynamically adjusting exploration rates and recognizing spatiotemporal patterns.

Proposes a new sampling method for online learning with cumulative oversampling.

The paper develops a reinforcement learning model to estimate ad impact considering delayed and cumulative effects.

The paper develops methods to predict the probability of achieving a user goal in a task, ensuring the system alerts when the probability falls below a threshold.

The paper tackles causal bandits for SEMs, proposing algorithms that avoid estimating $2^N$ reward distributions.

Trajectory-level supervision allows efficient offline reinforcement learning.

DSAC improves cooperative MARL with general utilities, converging faster than existing methods.

Thompson Sampling bounds for contextual bandits with sub-Gaussian rewards.

Contextual multi-armed bandit (MAB) algorithms have been shown promising for maximizing cumulative rewards in sequential decision tasks such as news article recommendation systems, web page ad placement algorithms, and mobile health. However, most of the proposed contextual MAB algorithms assume linear relationships be…

New method for RL with general utilities using variational policy gradient.

This work explains RL policies using causal models, revealing important patterns and failures.

The Multi-Armed Bandits (MAB) framework highlights the tension between acquiring new knowledge (Exploration) and leveraging available knowledge (Exploitation). In the classical MAB problem, a decision maker must choose an arm at each time step, upon which she receives a reward. The decision maker's objective is to maxi…