R3L: Relative Representations for Reinforcement Learning
Not on the map. Why?
Abstract (arXiv)
Visual Reinforcement Learning is a popular and powerful framework that takes full advantage of the Deep Learning breakthrough. It is known that variations in input domains (e.g., different panorama colors due to seasonal changes) or task domains (e.g., altering the target speed of a car) can disrupt agent performance, necessitating new training for each variation. Recent advancements in the field of representation learning have demonstrated the possibility of combining components from different neural networks to create new models in a zero-shot fashion. In this paper, we build upon relative representations, a framework that maps encoder embeddings to a universal space. We adapt this framework to the Visual Reinforcement Learning setting, allowing to combine agents components to create new agents capable of effectively handling novel visual-task pairs not encountered during training. Our findings highlight the potential for model reuse, significantly reducing the need for retraining and, consequently, the time and computational resources required.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 29 of its 45 references that are in the map. First posted on arXiv 19 Apr 2024.
Builds on
29 of 45 references.
References · 29
- 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
- 2015Continuous control with deep reinforcement learningcited by 3,279
- 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,401
- 2020Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixelscited by 573
- 2020Learning Invariant Representations for Reinforcement Learning without Reconstructioncited by 424
- 2021Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learningcited by 359
- 2016Successor Features for Transfer in Reinforcement Learningcited by 331
- 2020Transfer Learning in Deep Reinforcement Learning: A Surveycited by 259
- 2014Understanding image representations by measuring their equivariance and equivalencecited by 240
- 2019Recurrent Independent Mechanismscited by 219
- 2021Revisiting Model Stitching to Compare Neural Representationscited by 203
- 2022Relative representations enable zero-shot latent space communicationcited by 175
17 more
- 2020Generalization in Reinforcement Learning by Soft Data Augmentationcited by 132
- 2016Learning Modular Neural Network Policies for Multi-Task and Multi-Robot Transfercited by 124
- 2020Self-Supervised Policy Adaptation during Deploymentcited by 121
- 2021Stabilizing Deep Q-Learning with ConvNets and Vision Transformers under Data Augmentationcited by 101
- 2021Similarity and Matching of Neural Network Representationscited by 77
- 2023Latent Space Translation via Semantic Alignmentcited by 46
- 2022ASIF: Coupled Data Turns Unimodal Models to Multimodal Without Trainingcited by 45
- 2022Modular Lifelong Reinforcement Learning via Neural Compositioncited by 39
- 2018Natural Environment Benchmarks for Reinforcement Learningcited by 32
- 2017Robust and Efficient Transfer Learning with Hidden-Parameter Markov Decision Processescited by 28
- 2021Measuring Sample Efficiency and Generalization in Reinforcement Learning Benchmarks: NeurIPS 2020 Procgen Benchmarkcited by 23
- 2023From Bricks to Bridges: Product of Invariances to Enhance Latent Space Communicationcited by 20
- 2021Invariance Through Latent Alignmentcited by 8
- 2020Beyond Tabula-Rasa: a Modular Reinforcement Learning Approach for Physically Embedded 3D Sokobancited by 7
- 2020Adversarial Skill Learning for Robust Manipulationcited by 6
- 2021Learning without Knowing: Unobserved Context in Continuous Transfer Reinforcement Learningcited by 4
- 2023Policy Stitching: Learning Transferable Robot Policiescited by 4
Built on by
Citing papers per year: 2026: 2.
2 papers.