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-free reinforcement learning has been successfully applied to a range ofchallenging problems, and has recently been extended to handle large neuralnetwork policies and value functions. However, the sample complexity ofmodel-free algorithms, particularly when using high-dimensional functionapproximators, tends to limit their applicability to physical systems. In thispaper, we explore algorithms and representations to reduce the samplecomplexity of deep reinforcement learning for continuous control tasks. Wepropose two complementary techniques for improving the efficiency of suchalgorithms. First, we derive a continuous variant of the Q-learning algorithm,which we call normalized adantage functions (NAF), as an alternative to themore commonly used policy gradient and actor-critic methods. NAF representationallows us to apply Q-learning with experience replay to continuous tasks, andsubstantially improves performance on a set of simulated robotic control tasks.To further improve the efficiency of our approach, we explore the use oflearned models for accelerating model-free reinforcement learning. We show thatiteratively refitted local linear models are especially effective for this, anddemonstrate substantially faster learning on domains where such models areapplicable.

Continuous Deep Q-Learning with Model-based Acceleration