Lecture 4B (2022-02-17): Niching Methods for Multimodal Optimization

This lecture concludes our coverage of genetic algorithms for multi-modal optimization, with a particular focus on niche-preserving methods within multi-modal genetic algorithms. We cover generalized fitness sharing approaches, clearing approaches, clustering approaches, crowding and tournament selection approaches (including restricted tournament selection, RTS), as well as conservation based approaches (like the species preserving GA, SPGA). Each of these methods helps to maximize diversity over large scales in genotypic space while still providing optimization/selective pressure at smaller scales -- providing both exploration and exploitation. They are inspired by ideas from conservation biology and community ecology, with some approaches borrowing from machine learning (as in K-means clustering).

Whiteboard lecture notes for this lecture can be found at: https://www.dropbox.com/s/kemycgsjb0t2hoc/IEE598-Lecture4B-2022-02-17-Niching_Methods_for_Multimodal_Optimization.pdf?dl=0