Component-wise Analysis of Automatically Designed Multiobjective Algorithms on Constrained Problems

TL;DR

This paper introduces a methodology to analyze the influence of components in automatically designed multiobjective algorithms, revealing that restart and update strategies significantly impact performance depending on problem difficulty.

Contribution

It proposes a new approach to investigate component effects in auto-designed algorithms and applies it to analyze a MOEA/D variant on constrained problems.

Findings

Restart and update strategies are the most influential components.

Their impact varies with problem difficulty.

Strategies improve performance and diversity metrics.

Abstract

The performance of multiobjective algorithms varies across problems, making it hard to develop new algorithms or apply existing ones to new problems. To simplify the development and application of new multiobjective algorithms, there has been an increasing interest in their automatic design from component parts. These automatically designed metaheuristics can outperform their human-developed counterparts. However, it is still uncertain what are the most influential components leading to their performance improvement. This study introduces a new methodology to investigate the effects of the final configuration of an automatically designed algorithm. We apply this methodology to a well-performing Multiobjective Evolutionary Algorithm Based on Decomposition (MOEA/D) designed by the irace package on nine constrained problems. We then contrast the impact of the algorithm components in terms of their Search Trajectory Networks (STNs), the diversity of the population, and the hypervolume. Our results indicate that the most influential components were the restart and update strategies, with higher increments in performance and more distinct metric values. Also, their relative influence depends on the problem difficulty: not using the restart strategy was more influential in problems where MOEA/D performs better; while the update strategy was more influential in problems where MOEA/D performs the worst.