Reward-Free RL

RL: Reinforcement Learning

Reward-free reinforcement learning asks how an agent can learn an environment and gather useful experience before knowing which reward or task it will face. Methods include intrinsic-motivation objectives such as entropy and novelty, uncertainty-driven exploration, and model-based or occupancy-coverage planning; learned representations and world models can make this data reusable. Research evaluates these approaches on sparse-reward navigation, continuous control, downstream task adaptation, and safe transfer, alongside theoretical sample-complexity bounds for finite Markov decision processes. A central challenge is balancing broad coverage with information that helps future tasks, while making exploration scalable and reliable beyond tabular settings.

Latest papers

16 papers4 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

DT-JEPA: Learning to Predict Future Representations for Decision-Making
Oct 23, 2026Changxu Ni, Yiyang Ni, Jun Li +4Transformer-Based RL Self-Supervised RL
Southeast University · Jiangsu Second Normal University · The University of Hong Kong +1
Self-organizing reinforcement learning: a learner-free swarm approaches the explore-exploit optimum
Oct 18, 2026Galen J. WilkersonMulti-Agent Resource Allocation Exploration-Exploitation Tradeoff
Komplexity AI LLC
Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction
Oct 14, 2026Robin Uhrich, Marco Bagatella, Núria Armengol Urpí +2Reinforcement Learning Policy Learning
University of Tübingen · ETH Zürich · Max Planck Institute for Intelligent Systems
Deep Epistemic Value Functions for Optimistic Exploration
Sep 28, 2026Leander Diaz-Bone, Marco Bagatella, Jonas Hübotter +1Value Function Estimation RL Exploration
ETH Zürich, Switzerland · Max Planck Institute for Intelligent Systems, Germany
Provably Safe Sim-to-Real Transfer
Sep 1, 2026Tingting Ni, Maryam KamgarpourReinforcement Learning Sim-to-Real Transfer
SYCAMORE, EPFL
Reward-Free Continual Adaptation for Resilient Space Robots
Aug 24, 2026Andrej Orsula, Miguel Olivares-Mendez, Carol MartinezReinforcement Learning Continual Robot Learning
University of Luxembourg
SP3O: Reinforcement Learning from Segment Preferences without Reward Modeling
Aug 3, 2026Evan Assmus, Qining Zhang, Lei YingReinforcement Learning Robotic RL
University of Michigan, Ann Arbor
Reward-free Pretraining for Reinforcement Learning via Occupancy Coverage Maximization
Jun 19, 2026Marco Pratticò, Pietro Novelli, Massimiliano Pontil +1Reinforcement Learning Exploration-Exploitation Tradeoff
Computational Statistics and Machine Learning - Istituto Italiano di Tecnologia, 16100 Genova, Italy · 2AI Centre, Computer Science Department, University College London, London, UK.
Online Reward-Punishment Learning from Fixed-Channel Perceptual Event Streams without Environment Rewards
Jun 17, 2026Zirong LiReward Modeling Reinforcement Learning
Tiangong University
Label-Free Reinforcement Learning via Cross-Model Entropy
May 27, 2026Matt Gorbett, Hossein ShiraziReinforcement Learning Instruction Following
Independent Researcher · San Diego State University
A Reward-Free Viewpoint on Multi-Objective Reinforcement Learning
Apr 27, 2026Ying-Tu Chen, Wei Hung, Bing-Shu Wu +2Multi-Objective Reinforcement Learning Reinforcement Learning
National Yang Ming Chiao Tung University, Hsinchu, Taiwan · Massachusetts Institute of Technology
Maximum Entropy Exploration Without the Rollouts
Mar 12, 2026Jacob Adamczyk, Adam Kamoski, Rahul V. KulkarniRL Exploration Maximum Entropy RL
Department of Physics, University of Massachusetts Boston · NSF Institute for Artificial Intelligence and Fundamental Interactions
Temporal Representations for Exploration: Learning Complex Exploratory Behavior without Extrinsic Rewards
Mar 2, 2026Faisal Mohamed, Catherine Ji, Benjamin Eysenbach +1RL Exploration Robotic RL
Mila-Quebec AI Institute · Université de Montréal · Princeton University
Improved Bounds for Reward-Agnostic and Reward-Free Exploration
Feb 18, 2026Oran Ridel, Alon CohenRL Exploration Reward-Free RL
Department of Engineering, Tel Aviv University, Tel Aviv, Israel. · Google Research, Tel Aviv, Israel.
A Unified Framework for Zero-Shot Reinforcement Learning
Oct 23, 2025Jacopo Di Ventura, Jan Felix Kleuker, Aske Plaat +1Zero-Shot RL Formal Methods for RL
Leiden University
ELEMENT: Episodic and Lifelong Exploration via Maximum Entropy
Dec 5, 2024Hongming Li, Zhao Yang, Xiaoxuan Liang +2Reward-Free RL Reinforcement Learning
Zhejiang Lab, Huangzhou, China · Vrije Universiteit Amsterdam, Amsterdam, the Netherlands · Computational NeuroEngineering Laboratory at the University of Florida, Gainesville, USA

Oct 23, 2026J/K move · Enter open · S save

DT-JEPA: Learning to Predict Future Representations for Decision-Making

Changxu Ni, Yiyang Ni, Jun Li, Feiran You+3

Southeast University · Jiangsu Second Normal University · The University of Hong Kong+1

Transformer-Based RL Self-Supervised RL Predictive Representation Learning Offline-To-Online RL Joint-Embedding Predictive Architecture Reward-Free RL

Open paper AskPDF arXiv

Decision Transformer (DT) has demonstrated strong performance in various offline reinforcement learning (RL) tasks by conditioning action generation on expected future returns and historical trajectories. However, treating expected future returns as the condition provides an incomplete description of the future, as trajectories with different state transitions and behaviors may share similar returns. Moreover, expected future returns inherently depend on task-specific reward functions, which makes policy learning reliant on reward-labeled datasets and restricts its scalability to tasks with similar but unseen rewards. Recently, Joint Embedding Predictive Architectures (JEPA) have emerged as a promising paradigm of self-supervised learning-based world models, which learn predictive representations of future dynamics in latent space rather than reconstructing future states in the original space. Inspired by JEPA, we propose DT-JEPA, an offline pretraining and online fine-tuning framework that learns general knowledge from reward-free offline trajectories and achieves excellent adaptation to unseen tasks. Instead of predicting a single deterministic future embedding, we design a novel probabilistic predictor to model the conditional distribution of future representations in the latent space, thereby capturing both predictive information and uncertainty about future dynamics. Theoretically, we show that 1) the JEPA objective seeks to maximize a variational lower bound on the mutual information between historical representations and future trajectories; and 2) predictive future-conditioned policy learning admits a finite-sample performance bound with an O(N-1/4) estimation term. Experiments on D4RL benchmarks demonstrate that DT-JEPA achieves efficient adaptation across varying offline RL tasks, outperforming traditional DT-based and other offline RL methods.