Zero-shot Generalization Of Vision-based RL Without Data Augmentation

Abstract

Generalizing vision-based reinforcement learning (RL) agents to novel environments remains a difficult and open challenge. Current trends are to collect large-scale datasets or use data augmentation techniques to prevent overfitting and improve downstream generalization. However, the computational and data collection costs increase exponentially with the number of task variations and can destabilize the already difficult task of training RL agents. In this work, we take inspiration from recent advances in computational neuroscience and propose a model, Associative Latent DisentAnglement (ALDA), that builds on standard off-policy RL towards zero-shot generalization. Specifically, we revisit the role of latent disentanglement in RL and show how combining it with a model of associative memory achieves zero-shot generalization on difficult task variations without relying on data augmentation. Finally, we formally show that data augmentation techniques are a form of weak disentanglement and

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keybatra2024zero

Related papers

• Temporal Disentanglement Of Representations For Improved Generalisation In Reinforcement Learning (2022)0.00
• Zero-shot Policy Learning With Spatial Temporal Rewarddecomposition On Contingency-aware Observation (2019)0.00
• A Recipe For Unbounded Data Augmentation In Visual Reinforcement Learning (2024)0.00
• Generalization Of Reinforcement Learning With Policy-aware Adversarial Data Augmentation (2021)0.00
• DARLA: Improving Zero-shot Transfer In Reinforcement Learning (2017)0.00
• Cross-trajectory Representation Learning For Zero-shot Generalization In RL (2021)0.00
• Inferring Behavior-specific Context Improves Zero-shot Generalization In Reinforcement Learning (2024)0.95
• DRED: Zero-shot Transfer In Reinforcement Learning Via Data-regularised Environment Design (2024)1.81