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

Developments in reinforcement learning (RL) have allowed algorithms toachieve impressive performance in highly complex, but largely static problems.In contrast, biological learning seems to value efficiency of adaptation to aconstantly-changing world. Here we build on a recently-proposed neuronallearning rule that assumes each neuron can optimize its energy balance bypredicting its own future activity. That assumption leads to a neuronallearning rule that uses presynaptic input to modulate prediction error. Weargue that an analogous RL rule would use action probability to modulate rewardprediction error. This modulation makes the agent more sensitive to negativeexperiences, and more careful in forming preferences. We embed the proposedrule in both tabular and deep-Q-network RL algorithms, and find that itoutperforms conventional algorithms in simple, but highly-dynamic tasks. Wesuggest that the new rule encapsulates a core principle of biologicalintelligence; an important component for allowing algorithms to adapt to changein a human-like way.

Reinforcement Learning with Brain-Inspired Modulation can Improve Adaptation to Environmental Changes