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

Exploration in reinforcement learning is a challenging problem: in the worstcase, the agent must search for high-reward states that could be hiddenanywhere in the state space. Can we define a more tractable class of RLproblems, where the agent is provided with examples of successful outcomes? Inthis problem setting, the reward function can be obtained automatically bytraining a classifier to categorize states as successful or not. If trainedproperly, such a classifier can provide a well-shaped objective landscape thatboth promotes progress toward good states and provides a calibrated explorationbonus. In this work, we show that an uncertainty aware classifier can solvechallenging reinforcement learning problems by both encouraging exploration andprovided directed guidance towards positive outcomes. We propose a novelmechanism for obtaining these calibrated, uncertainty-aware classifiers basedon an amortized technique for computing the normalized maximum likelihood (NML)distribution. To make this tractable, we propose a novel method for computingthe NML distribution by using meta-learning. We show that the resultingalgorithm has a number of intriguing connections to both count-basedexploration methods and prior algorithms for learning reward functions, whilealso providing more effective guidance towards the goal. We demonstrate thatour algorithm solves a number of challenging navigation and roboticmanipulation tasks which prove difficult or impossible for prior methods.

MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning