def init_population(self, X, y):
n_x, n_y = X.shape[0], y.shape[0]
population = []
for i in range(self.pop_size):
c = Chromosome(n_x, n_y)
f = c.compute_fitness(X, y, self.activation)
c.set_fitness(f)
population.append(c)
if f > self.best_fitness:
self.best_fitness = f
self.best_chromosome = copy.deepcopy(c)
self.population = population
FUNCTION_ADF = FUNCTION + ADF
# read data
nd, nv, data = read_data(
"dataset/F0_1_training_data.txt") # "./dataset/F49_2_training_data.txt"
cuda = torch.device("cuda")
data = torch.tensor(data)
population = []
for _ in range(NP):
# random initialization
c = Chromosome(var_num=nv)
# evaluate initial population
c.compute_fitness(data)
population.append(c)
for epo in range(1000000):
# population.sort(key=lambda x: x.fitness)
# best_individual = population[0]
best_individual = sorted(population, key=lambda x: x.fitness)[0]
if epo % 10 == 0:
print("epoch:", epo, " fitness:", best_individual.fitness)
print(best_individual.gene)
if best_individual.fitness < 1e-5:
break