paper-with-me

홈 › Papers

Unsupervised Curricula for Visual Meta-Reinforcement Learning

2019-12-09 · NeurIPS 2019 12 · Allan Jabri, Kyle Hsu, Ben Eysenbach, Abhishek Gupta, Sergey Levine, Chelsea Finn

In principle, meta-reinforcement learning algorithms leverage experience across many tasks to learn fast reinforcement learning (RL) strategies that transfer to similar tasks. However, current meta-RL approaches rely on manually-defined distributions of training tasks, and hand-crafting these task distributions can be challenging and time-consuming. Can "useful" pre-training tasks be discovered in an unsupervised manner? We develop an unsupervised algorithm for inducing an adaptive meta-training task distribution, i.e. an automatic curriculum, by modeling unsupervised interaction in a visual environment. The task distribution is scaffolded by a parametric density model of the meta-learner's trajectory distribution. We formulate unsupervised meta-RL as information maximization between a latent task variable and the meta-learner's data distribution, and describe a practical instantiation which alternates between integration of recent experience into the task distribution and meta-learning of the updated tasks. Repeating this procedure leads to iterative reorganization such that the curriculum adapts as the meta-learner's data distribution shifts. In particular, we show how discriminative clustering for visual representation can support trajectory-level task acquisition and exploration in domains with pixel observations, avoiding pitfalls of alternatives. In experiments on vision-based navigation and manipulation domains, we show that the algorithm allows for unsupervised meta-learning that transfers to downstream tasks specified by hand-crafted reward functions and serves as pre-training for more efficient supervised meta-learning of test task distributions.

📄 PDF Abstract BibTeX arXiv:1912.04226

Code (0)

등록된 구현이 없습니다.

Tasks

ClusteringMeta-LearningMeta Reinforcement Learningreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Methods 이 논문이 사용한 방법론

Test 설명 없음

Similar Papers 제목 키워드 기반

Beyond Fixed Tasks: Unsupervised Environment Design for Task-Level Pairs

2025-11-16 · Daniel Furelos-Blanco, Charles Pert, Frederik Kelbel, Alex F. Spies 외 arxiv

Training general agents to follow complex instructions (tasks) in intricate environments (levels) remains a core challenge in reinforcement learning. Random sampling of task-level pairs often produces unsolvable combinat…

Reinforcement Learning

Discovering General Reinforcement Learning Algorithms with Adversarial Environment Design

2023-10-04 · NeurIPS 2023 11 · Matthew Thomas Jackson, Minqi Jiang, Jack Parker-Holder, Risto Vuorio 외

The past decade has seen vast progress in deep reinforcement learning (RL) on the back of algorithms manually designed by human researchers. Recently, it has been shown that it is possible to meta-learn update rules, wit…

Deep Reinforcement LearningGeneral Reinforcement Learningreinforcement-learningReinforcement Learning+1

Unsupervised Reinforcement Learning of Transferable Meta-Skills for Embodied Navigation

2019-11-18 · CVPR 2020 6 · Juncheng Li, Xin Wang, Siliang Tang, Haizhou Shi 외

Visual navigation is a task of training an embodied agent by intelligently navigating to a target object (e.g., television) using only visual observations. A key challenge for current deep reinforcement learning models l…

Deep Reinforcement LearningObjectreinforcement-learningReinforcement Learning+3

Unsupervised Learning of Efficient Exploration: Pre-training Adaptive Policies via Self-Imposed Goals

2026-01-27 · Octavio Pappalardo arxiv

Unsupervised pre-training can equip reinforcement learning agents with prior knowledge and accelerate learning in downstream tasks. A promising direction, grounded in human development, investigates agents that learn by …

Unsupervised Pre-trainingReinforcement Learning

Curricular Subgoals for Inverse Reinforcement Learning

2023-06-14 · Shunyu Liu, Yunpeng Qing, Shuqi Xu, Hongyan Wu 외

Inverse Reinforcement Learning (IRL) aims to reconstruct the reward function from expert demonstrations to facilitate policy learning, and has demonstrated its remarkable success in imitation learning. To promote expert-…

Autonomous DrivingD4RLImitation Learningreinforcement-learning+1