Rethinking Selection in Generational Genetic Algorithms to Solve Combinatorial Optimization Problems: An Upper Bound-based Parent Selection Strategy for Recombination

TL;DR

This paper introduces a deterministic parent selection strategy called UBS for genetic algorithms, leveraging an upper bound-based approach and a knowledge transfer mechanism to improve search efficiency on NP-hard combinatorial problems.

Contribution

It proposes a novel UBS strategy that uses a multi-armed bandit framework and similarity-based knowledge transfer to enhance parent selection in GAs for complex optimization tasks.

Findings

UBS favors larger population variations between generations.

UBS searches for high-quality solutions faster than traditional methods.

Effective on team orienteering and quadratic assignment problems.

Abstract

Existing stochastic selection strategies for parent selection in generational GA help build genetic diversity and sustain exploration; however, it ignores the possibility of exploiting knowledge gained by the process to make informed decisions for parent selection, which can often lead to an inefficient search for large, challenging optimization problems. This work proposes a deterministic parent selection strategy for recombination in a generational GA setting called Upper Bound-based Parent Selection (UBS) to solve NP-hard combinatorial optimization problems. Specifically, as part of the UBS strategy, we formulate the parent selection problem using the MAB framework and a modified UCB1 algorithm to manage exploration and exploitation. Further, we provided a unique similarity-based approach for transferring knowledge of the search progress between generations to accelerate the search. To demonstrate the effectiveness of the proposed UBS strategy in comparison to traditional stochastic selection strategies, we conduct experimental studies on two NP-hard combinatorial optimization problems: team orienteering and quadratic assignment. Specifically, we first perform a characterization study to determine the potential of UBS and the best configuration for all the selection strategies involved. Next, we run experiments using these best configurations as part of the comparison study. The results from the characterization studies reveal that UBS, in most cases, favors larger variations among the population between generations. Next, the comparison studies reveal that UBS can effectively search for high-quality solutions faster than traditional stochastic selection strategies on challenging NP-hard combinatorial optimization problems under given experimental conditions.