Research map / Reinforcement learning / Visual and representation RL

Learning Robust Representations with Long-Term Information for Generalization in Visual Reinforcement Learning

ICLR 2025 · Reinforcement learning › Visual and representation RL · approach: Robust visual encoders for RL

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes learning robust visual representations for generalization in visual RL that explicitly retain long-term information relevant to sequential decision-making, rather than only robustness to visual distractions.

Problem

Generalization in visual reinforcement learning to test environments with unseen visual distractions; existing robust representation methods ignore the sequential decision-making objective and so lose critical long-term information.

Approach

Incorporates long-term information into robust representation learning for VRL so that learned representations support downstream sequential decision-making under unseen visual distractions, instead of optimizing robustness alone.

Closest papers by meaning

• DMC-VB: A Benchmark for Representation Learning for Control with Visual Distractors (NeurIPS 2024 · similarity 0.88)
• Towards Principled Representation Learning from Videos for Reinforcement Learning (ICLR 2024 · similarity 0.87)
• The Surprising Ineffectiveness of Pre-Trained Visual Representations for Model-Based Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
• Sliding Puzzles Gym: A Scalable Benchmark for State Representation in Visual Reinforcement Learning (ICML 2025 · similarity 0.87)
• Robust Visual Imitation Learning with Inverse Dynamics Representations (AAAI 2024 · similarity 0.87)
• VIGOR: Zero-Shot Visual Generalization via Latent-Space Consistency in Model-Based Reinforcement Learning (NeurIPS 2026 · similarity 0.86)