Abstract:The emergence of networked systems in various fields brings many complex distributed optimization problems, where multiple agents in the system need to optimize a global objective cooperatively when they only have local information. In this work, we take advantage of the intrinsic parallelism of evolutionary computation to address network-based distributed optimization. In the proposed multiagent co-evolutionary algorithm, each agent maintains a subpopulation in which individuals represent solutions to the problem. During optimization, agents perform local optimization on their subpopulations and negotiation through communication with their neighbors. In order to help agents optimize the global objective cooperatively, we design a penalty-based objective function for fitness evaluation, which constrains the subpopulation within a small and controllable range. Further, to make the penalty more targeted, a conflict detection method is proposed to examine whether agents are conflicting on a certain shared variable. Finally, in order to help agents negotiate a consensus solution when only the local objective function is known, we retrofit the processes of negotiating shared variables, namely, evaluation, competition, and sharing. The above approaches form a multiagent co-evolutionary framework, enabling agents to cooperatively optimize the global objective in a distributed manner. Empirical studies show that the proposed algorithm achieves comparable solution quality with the holistic algorithm and better performance than existing gradient-free distributed algorithms on gradient-uncomputable problems.

Multi-objective Optimization Model and Evolutional Solution of Network Node Matching Problem

Mathematic model of node matching based on adjacency matrix and evolutionary solutions

An New Evolutionary Multi-Objective Optimization Algorithm

Multi-objective Evolutionary Algorithm for Enhancing the Robustness of Networks

Network Structural Balance Based on Evolutionary Multiobjective Optimization: A Two-Step Approach

Network Models and Optimization: Multiobjective Genetic Algorithm Approach

A Multi-Objective Approach For Community Detection In Complex Network

An Evolutionary Multiobjective Framework for Complex Network Reconstruction Using Community Structure

An improved differential evolution algorithm for solving the optimization model of multi-objective network

A Computationally Efficient Evolutionary Algorithm for Multiobjective Network Robustness Optimization

A Simulation-Based Multi-Objective Genetic Algorithm Approach for Networked Enterprises Optimization.

Community Detection in Networks by Using Multiobjective Evolutionary Algorithm with Decomposition

Constructing Robust Cooperative Networks using a Multi-Objective Evolutionary Algorithm

Multi-Objective Optimization of Transport Processes on Complex Networks.

A Multiagent Co-Evolutionary Algorithm with Penalty-Based Objective for Network-Based Distributed Optimization

An Evolutionary Algorithm Based on Fully Connected Weight Networks for Mixed-Variable Multi-objective Optimization

Multiobjective Evolutionary Model of the Construction Industry Based on Network Planning

Overlapping Community Detection in Complex Networks Using Multi-Objective Evolutionary Algorithm

An Evolutionary Algorithm Based on Fully Connected Weight Networks for Mixed-Variable Multiobjective Optimization

A multi-agent complex network algorithm for multi-objective optimization

Survey on Application of Evolutionary Algorithms in Complex Networks