The Road to VEGAS: Guiding the Search over Neutral Networks
Marie-Eleonore Marmion (LIFL), Clarisse Dhaenens (LIFL, INRIA Lille -, Nord Europe), Laetitia Jourdan (LIFL, INRIA Lille - Nord Europe), Arnaud, Liefooghe (LIFL, INRIA Lille - Nord Europe), S\'ebastien Verel (INRIA Lille -, Nord Europe)

TL;DR
VEGAS introduces a novel search methodology that leverages the entire neutral network and evolvability metrics to effectively navigate plateaus in optimization problems, enhancing search efficiency.
Contribution
The paper presents VEGAS, a new adaptive search method that considers the whole neutral network and uses multi-armed bandits to improve exploration in neutral landscapes.
Findings
Considering the entire neutral network improves search performance.
Evolvability-guided strategies effectively escape plateaus.
The exploration-exploitation balance impacts search success.
Abstract
VEGAS (Varying Evolvability-Guided Adaptive Search) is a new methodology proposed to deal with the neutrality property of some optimization problems. ts main feature is to consider the whole neutral network rather than an arbitrary solution. Moreover, VEGAS is designed to escape from plateaus based on the evolvability of solution and a multi-armed bandit. Experiments are conducted on NK-landscapes with neutrality. Results show the importance of considering the whole neutral network and of guiding the search cleverly. The impact of the level of neutrality and of the exploration-exploitation trade-off are deeply analyzed.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
