Improved NSGA-II Based on a Novel Ranking Scheme

TL;DR

This paper introduces an improved version of NSGA-II that reduces computational complexity through a novel ranking scheme, enhancing its efficiency for large-scale multiobjective optimization problems.

Contribution

It proposes a new variant of NSGA-II that decreases runtime complexity by leveraging a space-time trade-off, improving performance on large populations.

Findings

The improved algorithm reduces runtime complexity.

It performs well on large population datasets.

Effective in classifying leukemia types from microarray data.

Abstract

Non-dominated Sorting Genetic Algorithm (NSGA) has established itself as a benchmark algorithm for Multiobjective Optimization. The determination of pareto-optimal solutions is the key to its success. However the basic algorithm suffers from a high order of complexity, which renders it less useful for practical applications. Among the variants of NSGA, several attempts have been made to reduce the complexity. Though successful in reducing the runtime complexity, there is scope for further improvements, especially considering that the populations involved are frequently of large size. We propose a variant which reduces the run-time complexity using the simple principle of space-time trade-off. The improved algorithm is applied to the problem of classifying types of leukemia based on microarray data. Results of comparative tests are presented showing that the improved algorithm performs well on large populations.