Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Particle Swarm Optimization based on Novelty Search

Mr. Rajesh Misra, Kumar S Ray

TL;DR

This work introduces NSPSO, a hybrid that couples Novelty Search with Particle Swarm Optimization to tackle premature convergence in complex, multimodal, and high-dimensional landscapes. Novelty Search identifies novel search regions via region-level novelty, stored in an archive, and Leader particles trigger BBPSO optimization within those regions, iterating until novelty dissipates. Empirical results across 17 benchmark functions in 10 and 30 dimensions show NSPSO achieving strong performance, particularly in multimodal scenarios, and demonstrating robustness against center-bias in advanced benchmarks. The study highlights how objective-free exploration complements exploitation in PSO, offering a practical framework for scalable, diverse search in challenging optimization problems.

Abstract

In this paper we propose a Particle Swarm Optimization algorithm combined with Novelty Search. Novelty Search finds novel place to search in the search domain and then Particle Swarm Optimization rigorously searches that area for global optimum solution. This method is never blocked in local optima because it is controlled by Novelty Search which is objective free. For those functions where there are many more local optima and second global optimum is far from true optimum, the present method works successfully. The present algorithm never stops until it searches entire search area. A series of experimental trials prove the robustness and effectiveness of the present algorithm on complex optimization test functions.

Particle Swarm Optimization based on Novelty Search

TL;DR

This work introduces NSPSO, a hybrid that couples Novelty Search with Particle Swarm Optimization to tackle premature convergence in complex, multimodal, and high-dimensional landscapes. Novelty Search identifies novel search regions via region-level novelty, stored in an archive, and Leader particles trigger BBPSO optimization within those regions, iterating until novelty dissipates. Empirical results across 17 benchmark functions in 10 and 30 dimensions show NSPSO achieving strong performance, particularly in multimodal scenarios, and demonstrating robustness against center-bias in advanced benchmarks. The study highlights how objective-free exploration complements exploitation in PSO, offering a practical framework for scalable, diverse search in challenging optimization problems.

Abstract

In this paper we propose a Particle Swarm Optimization algorithm combined with Novelty Search. Novelty Search finds novel place to search in the search domain and then Particle Swarm Optimization rigorously searches that area for global optimum solution. This method is never blocked in local optima because it is controlled by Novelty Search which is objective free. For those functions where there are many more local optima and second global optimum is far from true optimum, the present method works successfully. The present algorithm never stops until it searches entire search area. A series of experimental trials prove the robustness and effectiveness of the present algorithm on complex optimization test functions.
Paper Structure (27 sections, 50 equations, 11 figures, 5 tables)

This paper contains 27 sections, 50 equations, 11 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Brief Introduction to Novelty Search
  3. Basic Idea
  4. Novelty Search Algorithm
  5. Application Domain
  6. Novelty Search based PSO
  7. Basic concepts
  8. General framework
  9. Novelty of a Search Region
  10. Novelty Computation Procedure
  11. Novelty score calculation
  12. Novelty Threshold
  13. Variant of Particle Swarm Optimization used in our algorithm
  14. Algorithmic summary
  15. Experimental Result and Analysis
  16. ...and 12 more sections

Figures (11)

  • Figure 1: Circular region defined by Leader particle $P_i$ with radius r.
  • Figure 2: Randomly distributed five Leader particles $P_1,P_2,P_3,P_4,P_5$ with radius r in search domain.
  • Figure 3: Novelty Score checking with the leader particle ' s previous positions.
  • Figure 4: Novelty Score checking with other leader particle ' s. recompute message is sent to itself.
  • Figure 5: Distance between two regions $P_1$ and $P_2$ in a 2 dimensional search space.$d_1 = \sqrt{(X_2 -X_1)^2 - (Y_2 -Y_1)^2}$
  • ...and 6 more figures