Conservative data sharing for multi-task offline reinforcement learning
Offline reinforcement learning (RL) algorithms have shown promising results in domains
where abundant pre-collected data is available. However, prior methods focus on solving …
where abundant pre-collected data is available. However, prior methods focus on solving …
Usher: Unbiased sampling for hindsight experience replay
Dealing with sparse rewards is a long-standing challenge in reinforcement learning (RL).
Hindsight Experience Replay (HER) addresses this problem by reusing failed trajectories for …
Hindsight Experience Replay (HER) addresses this problem by reusing failed trajectories for …
Bias Resilient Multi-Step Off-Policy Goal-Conditioned Reinforcement Learning
L Wu, K Chen - arXiv preprint arXiv:2311.17565, 2023 - arxiv.org
In goal-conditioned reinforcement learning (GCRL), sparse rewards present significant
challenges, often obstructing efficient learning. Although multi-step GCRL can boost this …
challenges, often obstructing efficient learning. Although multi-step GCRL can boost this …
Data Sharing without Rewards in Multi-Task Offline Reinforcement Learning
Offline reinforcement learning (RL) bears the promise to learn effective control policies from
static datasets but is thus far unable to learn from large databases of heterogeneous …
static datasets but is thus far unable to learn from large databases of heterogeneous …
An unbiased method to train robots traveling in special conditions
T Zhou - AIP Conference Proceedings, 2024 - pubs.aip.org
It is a challenge to make robots move from one place to another on the shortest path and
avoid obstacles at the same time, especially when there are some special conditions occur …
avoid obstacles at the same time, especially when there are some special conditions occur …
[图书][B] Reinforcement Learning from Static Datasets: Algorithms, Analysis, and Applications
A Kumar - 2023 - search.proquest.com
Reinforcement learning (RL) provides a formalism for learning-based control. By attempting
to learn behavioral policies that can optimize a user-specified reward function, RL methods …
to learn behavioral policies that can optimize a user-specified reward function, RL methods …
[图书][B] Building Versatile Reinforcement Learning Agents with Offline Data
T Yu - 2022 - search.proquest.com
Recent advances in machine learning using deep neural networks have shown significant
successes in learning from large datasets. However, these successes concentrated on …
successes in learning from large datasets. However, these successes concentrated on …
[PDF][PDF] Multi-Task Offline Reinforcement Learning with Conservative Data Sharing
Many recent offline RL algorithms attain both good empirical performance and enjoy
theoretical guarantees, but their applicability is limited to settings where data is collected for …
theoretical guarantees, but their applicability is limited to settings where data is collected for …