New Evolutionary Algorithms for Determining Consensus of Ordered Partition Collectives

Abstract

The ordered partition structure is helpful when an expert has to classify elements of a set into given classes. Finding consensus for an ordered partition collective is very important in making decisions. A 2-Optimality (O2) consensus best represents a collective, and distances between it and collective elements are uniform. However, finding such consensus has yet to be widely examined for ordered partition collectives. The best algorithm for this task in the literature is the HG3 algorithm. This study proposed three evolutionary algorithms to solve this problem. The algorithm (μ, λ)-IES is developed based on (μ, λ)-ES. The IHG2 algorithm is developed by fuzing the local search, elitism strategy, duplicate elimination, dynamic crossover rate, and dynamic mutation rate. The IHG3 algorithm is increasing the balance of exploration and exploitation. This algorithm is developed by fuzing the longest-distance strategy (KLD), elitism strategy (IEBL), local search, duplicate elimination, dynamic crossover rate, and dynamic mutation rate. The simulation results show that these algorithms generate high consensus quality. The IHG3 algorithm provides consensus with the best quality in an acceptable running time.