Community Based Network Reconstruction for an Evolutionary Algorithm Framework

Abstract

Inferring complex and non-linear dynamic system using the data that is available plays an important role in many areas of work such as physical, social, biological and computer sciences. In order to address these issues, network structure using a number of evolutionary algorithms has been proposed. However, the important criteria like the community structure have been ignored while developing these methodologies. Accordingly, this proposed work is focused on developing a multi-objective network reconstruction based on community structure in order to improve the network construction using ES by boosting their reconstruction performance. This framework that is used to further improve their performance is known as the community-based framework. It is based on multi-objective metaheuristic algorithm that is based on population and can be used as the base optimizer. The original decision space of the community structure is divided using the proposed work. From the solution obtained, an improved solution using reduced decision space is implemented using the multi-objective evolutionary algorithm (MOEA). A test suite is also designed to verify the performance of community based network reconstruction with respect to the complex network issue. In the proposed reconstruction methodology based on community criteria, the MOEAs are incorporated and are used to bind the original version. A noticeable improvement is seen in the experimental results based on the proposed work on 30 reconstruction issues.