English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recently, diffusion model shines as a promising backbone for the sequencemodeling paradigm in offline reinforcement learning(RL). However, these worksmostly lack the generalization ability across tasks with reward or dynamicschange. To tackle this challenge, in this paper we propose a task-orientedconditioned diffusion planner for offline meta-RL(MetaDiffuser), whichconsiders the generalization problem as conditional trajectory generation taskwith contextual representation. The key is to learn a context conditioneddiffusion model which can generate task-oriented trajectories for planningacross diverse tasks. To enhance the dynamics consistency of the generatedtrajectories while encouraging trajectories to achieve high returns, we furtherdesign a dual-guided module in the sampling process of the diffusion model. Theproposed framework enjoys the robustness to the quality of collected warm-startdata from the testing task and the flexibility to incorporate with differenttask representation method. The experiment results on MuJoCo benchmarks showthat MetaDiffuser outperforms other strong offline meta-RL baselines,demonstrating the outstanding conditional generation ability of diffusionarchitecture.

MetaDiffuser: Diffusion Model as Conditional Planner for Offline Meta-RL