A Multiple-Output Program Tree Structure in Genetic Programming

This paper presents a new program tree structure in genetic programming which outputs multiple related values, hence serves as a more coherent multiclass classifier. The multiple outputting effect of the tree is achieved by making it simulate a kind of directed acyclic graph. The approach is examined and compared with the basic genetic programming approach on four multiclass object classification tasks with varying difficulty. The results show that the new approach greatly outperforms the basic genetic programming approach on all the tasks.