Enhanced Optimization with Composite Objectives and Novelty Pulsation

TL;DR

This paper introduces Novelty Pulsation, a method that alternates between novelty-based selection and local optimization in multi-objective search, leading to faster solutions and better generalization in deceptive problems.

Contribution

It proposes a novel systematic approach called novelty pulsation that enhances multi-objective search by balancing diversity and local optimization.

Findings

Achieved state-of-the-art solutions for sorting networks

Established a new world record for 20-line sorting network with 91 comparators

Discovered solutions with better generalization in stock trading

Abstract

An important benefit of multi-objective search is that it maintains a diverse population of candidates, which helps in deceptive problems in particular. Not all diversity is useful, however: candidates that optimize only one objective while ignoring others are rarely helpful. A recent solution is to replace the original objectives by their linear combinations, thus focusing the search on the most useful trade-offs between objectives. To compensate for the loss of diversity, this transformation is accompanied by a selection mechanism that favors novelty. This paper improves this approach further by introducing novelty pulsation, i.e. a systematic method to alternate between novelty selection and local optimization. In the highly deceptive problem of discovering minimal sorting networks, it finds state-of-the-art solutions significantly faster than before. In fact, our method so far has established a new world record for the 20-lines sorting network with 91 comparators. In the real-world problem of stock trading, it discovers solutions that generalize significantly better on unseen data. Composite Novelty Pulsation is therefore a promising approach to solving deceptive real-world problems through multi-objective optimization.