Modified Soft Brood Crossover in Genetic Programming
This work addresses a specific issue in evolutionary computation for researchers, but it is incremental as it builds upon existing crossover operators.
The authors tackled premature convergence in Genetic Programming for data modeling by proposing a modified soft brood crossover operator to improve population diversity, and they demonstrated its performance on three symbolic regression problems.
Premature convergence is one of the important issues while using Genetic Programming for data modeling. It can be avoided by improving population diversity. Intelligent genetic operators can help to improve the population diversity. Crossover is an important operator in Genetic Programming. So, we have analyzed number of intelligent crossover operators and proposed an algorithm with the modification of soft brood crossover operator. It will help to improve the population diversity and reduce the premature convergence. We have performed experiments on three different symbolic regression problems. Then we made the performance comparison of our proposed crossover (Modified Soft Brood Crossover) with the existing soft brood crossover and subtree crossover operators.