def ten_arr_test(file_name):
weights, topology = get_genotype_data_from_file(file_name + ".csv")
net = Network(topology)
net.set_weights(weights)
while True:
ins, out = get_test_ten_arr()
print("Answer: " + ", ".join("{:.2f}".format(x)
for x in net.process_input(ins)))
print(net)
class Agent:
def __init__(self, genotype, topology):
self.genotype = genotype
self.genotype.eval = []
self.output = []
self.network = Network(topology)
self.network.set_weights(genotype.params)
def __lt__(self, other):
return self.genotype.fitness < other.genotype.fitness
def evaluate(self, inputs, correct):
network_answer = self.network.process_input(inputs)
self.output = network_answer
evaluation = Agent.eval_diff(network_answer, correct)
self.genotype.eval.append(evaluation)
return evaluation
@staticmethod
def eval_diff(answer, correct):
diff_sum = 0
for i in range(len(answer)):
diff_sum = diff_sum + (answer[i] - correct[i])**2
return diff_sum
@staticmethod
def eval_nr_incorrect(answer, correct):
incorrect = 0
for i in range(len(answer)):
diff = correct[i] - answer[i]
if abs(diff) > 0.01:
incorrect = incorrect + 1
return incorrect
@staticmethod
def eval_progress(answer, correct):
sum_percent = 0
for i in range(len(answer)):
if correct[i] is 1:
sum_percent = sum_percent + answer[i]
else:
sum_percent = sum_percent + (1 - answer[i])
return sum_percent / len(answer)
