In this paper, we compare a basic linear genetic programming (LGP) algorithm against several LGP variants, proposed by us, on two sets of symbolic regression benchmarks. We evaluated the influence of methods to control bloat, investigated these techniques focused in growth of effective code, and examined an operator to consider two successful individuals as modules to be integrated into a new individual. Results suggest that methods that deal with program size, percentage of effective code, and subfunctions, can improve the quality of the final solutions.
Highlights:
- Linear Genetic Programming
- Solved symbolic regression problems (nguyen and keijzer)
- New variants
- Bloat control
- Effective code
Keywords:
进化算法。线性规划的遗传问题。线性回归。具有象征意义的回归。
進化演算法。線性規劃的遺傳問題。線性回歸。具有象徵意義的回歸。
algorithme évolutionnaire. programmation génétique linéaire. régression linéaire. régression symbolique.
Evolutionärer Algorithmus. lineare genetische Programmierung. lineare Regression. symbolische Regression.
विकासवादी एल्गोरिथ्म। रैखिक आनुवंशिक प्रोग्रामिंग। रेखीय प्रतिगमन। प्रतीकात्मक प्रतिगमन।
evolusi algoritma. pengaturcaraan linear genetik. regresi linear. simbolik regresi.
algoritmo evolutivo. programación genética lineal. regresión lineal. regresión simbólica.