Reproduce “Task-Agnostic Online Reinforcement Learning with an Infinite Mixture of Gaussian Processes” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mengdi Xu, Wenhao Ding, Jiacheng Zhu, Zuxin Liu, Baiming Chen, Ding Zhao
The authors present a model-based reinforcement-learning method that uses a mixture of Gaussian-process experts to handle changing dynamics without task labels or pre-training. A temporal transition prior assigns observations to experts, creates models for new dynamics, and recalls models for returning dynamics. In simulation, CartPole-SwingUp Task 2 reward exceeded a single-GP baseline; this does not establish real-world performance or gains across tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.