def main():
X, Y = load_data(sys.argv[1])
X_val, Y_val = load_data(sys.argv[2])
X_test, Y_test = load_data(sys.argv[3])
print("1", X.shape)
print("2", Y.shape)
red = NeuralNetwork()
red.type = 'C'
#1
red.structure = [5, 4, 5]
red.initialize_parameters()
red.back_propagation2(X, Y, 10, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(red.get_coded_y(predicts), red.get_coded_y(original))
print("--------------------")
#2
red.structure = [5, 16, 5]
red.initialize_parameters()
red.back_propagation2(X, Y, 10, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(red.get_coded_y(predicts), red.get_coded_y(original))
print("--------------------")
# 3
red.structure = [5, 32, 5]
red.initialize_parameters()
red.back_propagation2(X, Y, 10, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(red.get_coded_y(predicts), red.get_coded_y(original))
print("--------------------")
#4
red.structure = [5, 64, 5]
red.initialize_parameters()
red.back_propagation2(X, Y, 10, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(red.get_coded_y(predicts), red.get_coded_y(original))
print("--------------------")
#5
red.structure = [5, 128, 5]
red.initialize_parameters()
red.back_propagation2(X, Y, 10, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(red.get_coded_y(predicts), red.get_coded_y(original))
print("--------------------")
def main():
X, Y = load_data(sys.argv[1])
X_val, Y_val = load_data(sys.argv[2])
X_test, Y_test = load_data(sys.argv[3])
red = NeuralNetwork()
red.out = 'sigmoid'
#1
red.structure = [7, 16, 1]
red.initialize_parameters()
print("X", X.shape)
print("Y", Y.shape)
red.back_propagation3(X, Y, 50, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(predicts, original)
#2
red.structure = [7, 32, 1]
red.initialize_parameters()
print("X", X.shape)
print("Y", Y.shape)
red.back_propagation3(X, Y, 50, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(predicts, original)
#3
red.structure = [7, 16, 16, 1]
red.initialize_parameters()
print("X", X.shape)
print("Y", Y.shape)
red.back_propagation3(X, Y, 50, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(predicts, original)
#4
red.structure = [7, 32, 32, 1]
red.initialize_parameters()
print("X", X.shape)
print("Y", Y.shape)
red.back_propagation3(X, Y, 50, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(predicts, original)
#5
red.structure = [7, 4, 4, 4, 1]
red.initialize_parameters()
print("X", X.shape)
print("Y", Y.shape)
red.back_propagation3(X, Y, 50, X_val, Y_val, 0.05, 3)
predicts = []
original = []
X_test_t = np.transpose(X_test)
Y_test_t = np.transpose(Y_test)
for j in range(len(X_test_t)):
X_eval = X_test_t[j].reshape(len(X_test_t[j]), 1)
Y_eval = Y_test_t[j].reshape(len(Y_test_t[j]), 1)
y_predict, _ = red.L_model_forward(X_eval)
predicts.append(y_predict.reshape(len(y_predict)))
original.append(Y_eval.reshape(len(Y_eval)))
stadistics(predicts, original)
