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

Deep Reinforcement Learning (DRL) connects the classic Reinforcement Learningalgorithms with Deep Neural Networks. A problem in DRL is that CNNs areblack-boxes and it is hard to understand the decision-making process of agents.In order to be able to use RL agents in highly dangerous environments forhumans and machines, the developer needs a debugging tool to assure that theagent does what is expected. Currently, rewards are primarily used to interprethow well an agent is learning. However, this can lead to deceptive conclusionsif the agent receives more rewards by memorizing a policy and not learning torespond to the environment. In this work, it is shown that this problem can berecognized with the help of gradient visualization techniques. This work bringssome of the best-known visualization methods from the field of imageclassification to the area of Deep Reinforcement Learning. Furthermore, two newvisualization techniques have been developed, one of which providesparticularly good results. It is being proven to what extent the algorithms canbe used in the area of Reinforcement learning. Also, the question arises on howwell the DRL algorithms can be visualized across different environments withvarying visualization techniques.

Are Gradient-based Saliency Maps Useful in Deep Reinforcement Learning?