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/

Zendy Plus

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

Zendy Tools

Zendy Open

Modularity has been widely studied as a mechanism to improve the capabilitiesof neural networks through various techniques such as hand-crafted modulararchitectures and automatic approaches. While these methods have sometimesshown improvements towards generalisation ability, robustness, and efficiency,the mechanisms that enable modularity to give performance advantages areunclear. In this paper, we investigate this issue and find that the amount ofnetwork modularity for optimal performance is likely entangled in complexrelationships between many other features of the network and problemenvironment. Therefore, direct optimisation or arbitrary designation of asuitable amount of modularity in neural networks may not be beneficial. We useda classic neuroevolutionary algorithm which enables rich, automaticoptimisation and exploration of neural network architectures and weights withvarying levels of modularity. The structural modularity and performance ofnetworks generated by the NeuroEvolution of Augmenting Topologies algorithm wasassessed on three reinforcement learning tasks, with and without an additionalmodularity objective. The results of the quality-diversity optimisationalgorithm, MAP-Elites, suggest intricate conditional relationships betweenmodularity, performance, and other predefined network features.

Towards Understanding the Link Between Modularity and Performance in Neural Networks for Reinforcement Learning