Understanding Parameter Spaces Using Local Optima Networks

Explore the complexities of parameter spaces in metaheuristics through a conference talk focusing on Particle Swarm Optimization (PSO). Delve into the use of Local Optima Networks (LONs) to visualize and analyze PSO parameter landscapes across various objective functions. Discover how the underlying objective function influences the structure of parameter landscapes, revealing unexpected complexities in PSO parameter tuning. Gain insights into the formalism of parameter landscapes and learn how LONs serve as an effective tool for analyzing and visualizing metaheuristic parameter landscapes. Follow the presentation's structure, covering the introduction, parameter landscape overview, PSO specifics, LONs, experimental results, and key observations from functions like Egg Holder and HyperEllipsoid.