[keras] DNN (Basic CIFAR10)

 Source : <"Coding chef's 3 minute deep running kerasu taste" "Kim Sung-jin">

코드

# DNN_CIFAR-10 / MNIST보다 복잡한 데이터의 처리(R,G,B)
import numpy as np
from keras import datasets
from keras.utils import np_utils
from keras import layers, models


class DNN(models.Sequential):
    def __init__(self, Nin, Nh_l, Pd_l, Nout):
        super().__init__()

		# 첫 번째 은닉층
        self.add(layers.Dense(Nh_l[0], activation='relu',input_shape=(Nin,), name='Hidden-1'))
		# Dropout 확률을 정한다.
		# Dropout : 랜덤으로 몇개의 노드를 비활성화 한다.(오버피팅 방지)
        self.add(layers.Dropout(Pd_l[0]))
        self.add(layers.Dense(Nh_l[1], activation='relu', name='Hidden-2'))
        self.add(layers.Dropout(Pd_l[1]))
        self.add(layers.Dense(Nout, activation='softmax'))
        self.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
    
# 데이터 정리
def Data_func():
    (X_train, y_train), (X_test, y_test) = datasets.cifar10.load_data()

	# 0~9 사이의 정수값을 10개의 원소를 가지는 벡터로 변환
	# 1 -> 0100000000, 2 -> 0010000000, 9 -> 0000000001
    Y_train = np_utils.to_categorical(y_train)
    Y_test = np_utils.to_categorical(y_test)

    L, W, H, C = X_train.shape
    X_train = X_train.reshape(-1, W * H * C)
    X_test = X_test.reshape(-1, W * H * C)

    X_train = X_train / 255.0
    X_test = X_test / 255.0

    return (X_train, Y_train), (X_test, Y_test)

from ex_2_1_keras_ann import plot_acc, plot_loss
import matplotlib.pyplot as plt

# 모델 테스팅
def main():
    Nh_l = [100, 50]
    Pd_l = [0.0, 0.0]
    number_of_class = 10
    Nout = number_of_class

    (X_train, Y_train), (X_test, Y_test) = Data_func()
    model = DNN(X_train.shape[1], Nh_l, Pd_l, Nout)
    history = model.fit(X_train, Y_train, epochs=10, batch_size=100, validation_split=0.2)

    performace_test = model.evaluate(X_test, Y_test, batch_size=100)
    print('Test Loss and Accuracy ->', performace_test)

    plot_loss(history)
    plt.show()
    plot_acc(history)
    plt.show()


if __name__ == '__main__':
    main()

실행 결과