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

Designing evolutionary algorithms capable of uncovering highly evolvablerepresentations is an open challenge; such evolvability is important because itaccelerates evolution and enables fast adaptation to changing circumstances.This paper introduces evolvability ES, an evolutionary algorithm designed toexplicitly and efficiently optimize for evolvability, i.e. the ability tofurther adapt. The insight is that it is possible to derive a novel objectivein the spirit of natural evolution strategies that maximizes the diversity ofbehaviors exhibited when an individual is subject to random mutations, and thatefficiently scales with computation. Experiments in 2-D and 3-D locomotiontasks highlight the potential of evolvability ES to generate solutions withtens of thousands of parameters that can quickly be adapted to solve differenttasks and that can productively seed further evolution. We further highlight aconnection between evolvability and a recent and popular gradient-basedmeta-learning algorithm called MAML; results show that evolvability ES canperform competitively with MAML and that it discovers solutions with distinctproperties. The conclusion is that evolvability ES opens up novel researchdirections for studying and exploiting the potential of evolvablerepresentations for deep neural networks.

Evolvability ES: Scalable and Direct Optimization of Evolvability