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

Model-based reinforcement learning (RL) is considered to be a promisingapproach to reduce the sample complexity that hinders model-free RL. However,the theoretical understanding of such methods has been rather limited. Thispaper introduces a novel algorithmic framework for designing and analyzingmodel-based RL algorithms with theoretical guarantees. We design ameta-algorithm with a theoretical guarantee of monotone improvement to a localmaximum of the expected reward. The meta-algorithm iteratively builds a lowerbound of the expected reward based on the estimated dynamical model and sampletrajectories, and then maximizes the lower bound jointly over the policy andthe model. The framework extends the optimism-in-face-of-uncertainty principleto non-linear dynamical models in a way that requires \textit{no explicit}uncertainty quantification. Instantiating our framework with simplificationgives a variant of model-based RL algorithms Stochastic Lower BoundsOptimization (SLBO). Experiments demonstrate that SLBO achievesstate-of-the-art performance when only one million or fewer samples arepermitted on a range of continuous control benchmark tasks.

Algorithmic Framework for Model-based Deep Reinforcement Learning with Theoretical Guarantees