Illinois Genetic Algorithms Laboratory

In one of the GA session (GA-3) that I attended, two paper (entitled “Improvements to Penalty-Based Evolutionary Algorithms for the Multi-Dimensional Knapsack
Problem Using a Gene-Based Adaptive Mutation Approach” and “Adaptive Isolation Model using Data Clustering for Multimodal Function Optimization”) proposed a similar scheme of GAs: they both try to speed up the convergence, and then re-initialize the population to explore the search space. I think this trend leads to the issue about time continuation (for more information, please refer to [Goldberg, 1999]). The issue there is, should we use one epoch, large population, or multiple epochs, but smaller populations? Both the papers show promising results based on their test function; however, no theory or model is developed to support their results. I think we need more theories to guide us in this direction so we can better design the algorithms.