Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Socio-cognitive agent-oriented evolutionary algorithm with trust-based optimization

Aleksandra Urbańczyk, Krzysztof Czech, Piotr Urbańczyk, Marek Kisiel-Dorohinicki, Aleksander Byrski

TL;DR

The paper addresses premature convergence and inefficient information exchange in island-model evolutionary algorithms by introducing Trust-Based Optimization (TBO), which replaces periodic migrations with adaptive, trust- or reputation-based interactions among agent-populations. TBO employs a socio-cognitive crossover learning operator and an interaction protocol that shares high-quality solutions proportional to credibility, with diversity controlled by the factor $d_f$. Empirical results on six benchmark problems across multiple dimensions show that TBO often outperforms the baseline island model, though performance is configuration- and landscape-dependent. These findings demonstrate that socio-cognitive mechanisms can flexibly enhance convergence speed and solution quality in distributed EAs, with potential for heterogeneous-agent extensions and real-world applications.

Abstract

This paper introduces the Trust-Based Optimization (TBO), a novel extension of the island model in evolutionary computation that replaces conventional periodic migrations with a flexible, agent-driven interaction mechanism based on trust or reputation. Experimental results demonstrate that TBO generally outperforms the standard island model evolutionary algorithm across various optimization problems. Nevertheless, algorithm performance varies depending on the problem type, with certain configurations being more effective for specific landscapes or dimensions. The findings suggest that trust and reputation mechanisms provide a flexible and adaptive approach to evolutionary optimization, improving solution quality in many cases.

Socio-cognitive agent-oriented evolutionary algorithm with trust-based optimization

TL;DR

The paper addresses premature convergence and inefficient information exchange in island-model evolutionary algorithms by introducing Trust-Based Optimization (TBO), which replaces periodic migrations with adaptive, trust- or reputation-based interactions among agent-populations. TBO employs a socio-cognitive crossover learning operator and an interaction protocol that shares high-quality solutions proportional to credibility, with diversity controlled by the factor . Empirical results on six benchmark problems across multiple dimensions show that TBO often outperforms the baseline island model, though performance is configuration- and landscape-dependent. These findings demonstrate that socio-cognitive mechanisms can flexibly enhance convergence speed and solution quality in distributed EAs, with potential for heterogeneous-agent extensions and real-world applications.

Abstract

This paper introduces the Trust-Based Optimization (TBO), a novel extension of the island model in evolutionary computation that replaces conventional periodic migrations with a flexible, agent-driven interaction mechanism based on trust or reputation. Experimental results demonstrate that TBO generally outperforms the standard island model evolutionary algorithm across various optimization problems. Nevertheless, algorithm performance varies depending on the problem type, with certain configurations being more effective for specific landscapes or dimensions. The findings suggest that trust and reputation mechanisms provide a flexible and adaptive approach to evolutionary optimization, improving solution quality in many cases.

Paper Structure

This paper contains 14 sections, 17 equations, 2 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. The origin and motivation for the TBO
  3. TBO features
  4. Relations based on trust and reputation
  5. Learning mechanism---socio-cognitive crossover
  6. Interaction between agents
  7. TBO algorithm
  8. Experiments and results
  9. Baseline algorithm and parameters values
  10. Experiment setup
  11. Experimental results
  12. Conclusions
  13. Acknowledgments.
  14. Disclosure of Interests.

Figures (2)

  • Figure 1: Schema of the interaction step illustrated using a trust-based relationship.
  • Figure 2: Convergence analysis of Strong leadership (blue), Exploration (green), Small society (magenta), Large society (red), and High diversity (yellow) TBO configurations, as well as Island model (cyan) on six benchmark functions---200-di-men-sional Sphere (a), Griewank (b), Rastrigin (c), Expanded Shaffer (d), Schwefel with noise problem (e) evaluated over 400,000 iterations and 100-dimensional Lennard-Jones Minimum Energy Cluster problem (f) evaluated over 200,000 iterations.