Exploiting Tournament Selection for Efficient Parallel Genetic Programming

© 2020, Springer Nature Switzerland AG. Dynamic flexible job shop scheduling (DFJSS) has been widely studied in both academia and industry. Both machine assignment and operation sequencing decisions need to be made simultaneously as an operation can be processed by a set of machines in DFJSS. Using scheduling heuristics to solve the DFJSS problems becomes an effective way due to its efficiency and simplicity. Genetic programming (GP) has been successfully applied to evolve scheduling heuristics for job shop scheduling automatically. However, the subtrees of the selected parents are randomly chosen in traditional GP for crossover and mutation, which may not be sufficiently effective, especially in a huge search space. This paper proposes new strategies to guide the subtree selection rather than picking them randomly. To be specific, the occurrences of features are used to measure the importance of each subtree of the selected parents. The probability to select a subtree is based on its importance and the type of genetic operators. This paper examines the proposed algorithm on six DFJSS scenarios. The results show that the proposed GP algorithm with the guided subtree selection for crossover can converge faster and achieve significantly better performance than its counterpart in half of the scenarios while no worse in all other scenarios without increasing the computational time.

Download Full-text

Adaptive Test Selection for Factorization-based Surrogate Fitness in Genetic Programming

Foundations of Computing and Decision Sciences ◽

10.1515/fcds-2017-0017 ◽

2017 ◽

Vol 42 (4) ◽

pp. 339-358 ◽

Cited By ~ 1

Author(s):

Krzysztof Krawiec ◽

Paweł Liskowski

Keyword(s):

Genetic Programming ◽

Evolutionary Algorithm ◽

Matrix Factorization ◽

Fitness Function ◽

Interaction Matrix ◽

Simulated Evolution ◽

Adaptive Test ◽

Selection For ◽

Correct Program ◽

Non Negative Matrix Factorization

Abstract Genetic programming (GP) is a variant of evolutionary algorithm where the entities undergoing simulated evolution are computer programs. A fitness function in GP is usually based on a set of tests, each of which defines the desired output a correct program should return for an exemplary input. The outcomes of interactions between programs and tests in GP can be represented as an interaction matrix, with rows corresponding to programs in the current population and columns corresponding to tests. In previous work, we proposed SFIMX, a method that performs only a fraction of interactions and employs non-negative matrix factorization to estimate the outcomes of remaining ones, shortening GP’s runtime. In this paper, we build upon that work and propose three extensions of SFIMX, in which the subset of tests drawn to perform interactions is selected with respect to test difficulty. The conducted experiment indicates that the proposed extensions surpass the original SFIMX on a suite of discrete GP benchmarks.

Download Full-text

The unique implications of brood selection for genetic programming

Proceedings of the First IEEE Conference on Evolutionary Computation. IEEE World Congress on Computational Intelligence ◽

10.1109/icec.1994.350023 ◽

2002 ◽

Cited By ~ 5

Author(s):

W.A. Tackett ◽

A. Carmi

Keyword(s):

Genetic Programming ◽

Selection For

Download Full-text