A Brief Look at Generalization in Visual Meta-Reinforcement Learning

2020-06-12 · ICML Workshop LifelongML 2020 7 · Safa Alver, Doina Precup

Due to the realization that deep reinforcement learning algorithms trained on high-dimensional tasks can strongly overfit to their training environments, there have been several studies that investigated the generalization performance of these algorithms. However, there has been no similar study that evaluated the generalization performance of algorithms that were specifically designed for generalization, i.e. meta-reinforcement learning algorithms. In this paper, we assess the generalization performance of these algorithms by leveraging high-dimensional, procedurally generated environments. We find that these algorithms can display strong overfitting when they are evaluated on challenging tasks. We also observe that scalability to high-dimensional tasks with sparse rewards remains a significant problem among many of the current meta-reinforcement learning algorithms. With these results, we highlight the need for developing meta-reinforcement learning algorithms that can both generalize and scale.

📄 PDF Abstract BibTeX arXiv:2006.07262

Code (0)

등록된 구현이 없습니다.

Tasks

Deep Reinforcement LearningMeta Reinforcement Learningreinforcement-learning Reinforcement Learning Reinforcement Learning (RL)

A Brief Survey of Associations Between Meta-Learning and General AI

2021-01-12 · Huimin Peng

This paper briefly reviews the history of meta-learning and describes its contribution to general AI. Meta-learning improves model generalization capacity and devises general algorithms applicable to both in-distribution…

Meta-Learning

Look where you look! Saliency-guided Q-networks for generalization in visual Reinforcement Learning

2022-09-16 · David Bertoin, Adil Zouitine, Mehdi Zouitine, Emmanuel Rachelson

Deep reinforcement learning policies, despite their outstanding efficiency in simulated visual control tasks, have shown disappointing ability to generalize across disturbances in the input training images. Changes in im…

Deep Reinforcement Learning reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)

Metabolic graphs, LIFE method and the modeling of drug action on Mycobacterium tuberculosis

2020-03-26 · Sean T. McQuade, Nathaniel J. Merrill, Benedetto Piccoli

This paper serves as a framework for designing advanced models for drug action on metabolism. Drug treatment may affect metabolism by either enhancing or inhibiting metabolic reactions comprising a metabolic network. We …

Context Shift Reduction for Offline Meta-Reinforcement Learning

2023-11-07 · NeurIPS 2023 11 · Yunkai Gao, Rui Zhang, Jiaming Guo, Fan Wu 외

Offline meta-reinforcement learning (OMRL) utilizes pre-collected offline datasets to enhance the agent's generalization ability on unseen tasks. However, the context shift problem arises due to the distribution discrepa…

Meta Reinforcement Learning reinforcement-learning Reinforcement Learning Representation Learning

A Brief Summary of Interactions Between Meta-Learning and Self-Supervised Learning

2021-03-01 · Huimin Peng

This paper briefly reviews the connections between meta-learning and self-supervised learning. Meta-learning can be applied to improve model generalization capability and to construct general AI algorithms. Self-supervis…

Data Augmentation Meta-Learning Self-Supervised Learning Unsupervised Pre-training