Something went wrong. Try again.
Old project about training neuronal networks with genetic algorithms.
Something went wrong. Try again.
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137from scipy.stats.distributions import norm
__author__="Marcos Gabarda"__date__ ="$07-dic-2010 18:47:47$"
class EESEnvironment: """Extended Evolutionary Strategy
Environment to run extended evolutionary strategies.
""" def __init__(self, data_set, individual, initial_population=10, \ generations=100): self.__population = [] self.__parents = [] self.__offspring = [] self.__data_set = data_set self.__individual = individual self.__initial_population = initial_population self.__id_count = 0 self.__generations = generations if self.__data_set.mode == "cls": self.__problem = "max" else: self.__problem = "min" self.__selection_mode = "ml"
def __initialize(self): """ Initialize the initial population. """ for i in range(self.__initial_population): ind = object.__new__(self.__individual) ind.__init__( self.__data_set , i) ind.update_score() print ind self.__population.append(ind) self.__id_count = self.__initial_population
def __selection(self): """ Selection Phase.
Select all the population to be parents
""" self.__parents = [] for i in range(len(self.__population)): self.__parents.append(self.__population[i])
def __mutation(self, factor=7): """ Mutation Phase. """ self.__offspring = [] offspring_star = norm.rvs(loc=factor*len(self.__parents)) total_fitness = 0.0 for i in range(len(self.__parents)): total_fitness += self.__parents[i].score for i in range(len(self.__parents)): p = self.__parents[i] offspring_factor = 7 if total_fitness != 0.0: if self.__problem == "max": offspring_factor = offspring_star * (p.score / total_fitness) else: offspring_factor = offspring_star * (1-(p.score / total_fitness)) for j in range(int(offspring_factor)): self.__id_count += 1 o = p.mutate() o.id = self.__id_count o.update_score() self.__offspring.append(o)
def __replacement(self, mode="ml"): self.__population = [] individuals = [] if mode == "ml": for i in range(len(self.__offspring)): individuals.append((self.__offspring[i].score, self.__offspring[i])) else: individuals = [] for i in range(len(self.__offspring)): individuals.append((self.__offspring[i].score, self.__offspring[i])) for i in range(len(self.__parents)): individuals.append((self.__parents[i].score, self.__parents[i])) if self.__problem == "max": individuals = sorted(individuals, key=lambda ind: ind[0], reverse=True) else: individuals = sorted(individuals, key=lambda ind: ind[0]) for i in range(len(self.__parents)): self.__population.append(individuals[i][1]) self.__parents = [] self.__offspring = []
def __evolutionary_cycle(self): print "--- Selection..." self.__selection() print "--- Mutation..." self.__mutation() print "--- Replacement..." self.__replacement(mode=self.__selection_mode)
def get_best(self): individuals = [] if self.__problem == "max": individuals = sorted(self.__population, key=lambda ind: ind.score,\ reverse=True) else: individuals = sorted(self.__population, key=lambda ind: ind.score) return individuals[0]
def run(self): print "--- Initial population ---" self.__initialize() current_generation = 0 best_individual = None while current_generation < self.__generations: current_generation += 1 print "Generation " + str(current_generation) self.__evolutionary_cycle() best = self.get_best() if best_individual is None: best_individual = best else: if self.__problem == "max" and best.score > best_individual.score: best_individual = best elif self.__problem == "min" and best.score < best_individual.score: best_individual = best print "Best: " + str(best) print "(Parent ID: " + str(best.parent.id) + ")" return best_individual
if __name__ == "__main__": pass