FCN - tensorflow

import tensorflow as tf

import numpy as np

import pandas as pd

from keras.datasets.mnist import load_data

from keras.models import Sequential, Model

from keras.layers import Dense, Input ,Flatten, Dropout

from keras.utils import plot_model, to_categorical

from keras.regularizers import l2

from sklearn.model_selection import train_test_split

import matplotlib.pyplot as plt

from google.colab import drive

drive.mount('/content/drive')

tf.debugging.set_log_device_placement (False)

train_data = pd.read_csv('/content/drive/MyDrive/Colab Notebooks/fashion-mnist_train.csv')

test_data = pd.read_csv('/content/drive/MyDrive/Colab Notebooks/fashion-mnist_test.csv')

#tf.debugging.set_log_device_placement(True)

gpus = tf.config.list_physical_devices('GPU')

if gpus:

  # 텐서플로가 첫 번째 GPU만 사용하도록 제한

  try:

    tf.config.set_visible_devices(gpus[0], 'GPU')

  except RuntimeError as e:

    # 프로그램 시작시에 접근 가능한 장치가 설정되어야만 합니다

    print(e)

train_data = np.array(train_data)

test_data = np.array(test_data)

x = train_data[:,1:]

y =  train_data[:,0]

x_test = test_data[:,1:]

y_test = test_data[:,0]

x = x/255

x_test = x_test / 255

y = to_categorical(y)

y_test = to_categorical(y_test)

input1 = Input(shape = 784)

f=Dense(392,activation = 'relu')(input1)

f = Dropout (0.1)(f)

f=Dense(392,activation = 'relu')(f)

f = Dense(181,activation = 'relu', kernel_regularizer = l2(0.01))(f)

f=Dense(90,activation = 'relu')(f)

f = Dense(40,activation = 'relu', kernel_regularizer = l2(0.01))(f)

output=Dense(10,activation = 'softmax')(f)

model2 = Model(inputs = input1 , outputs = output)

plot_model(model2,show_shapes = True)

model2.compile(optimizer='adam',loss = 'categorical_crossentropy',metrics=['acc'])

hist = model2.fit(x,y,epochs = 80, batch_size = 500, validation_split = 0.3)

plt.subplot(1,2,1)

plt.plot(hist.history['loss'],label = 'train loss')

plt.plot(hist.history['val_loss'], label = 'val loss')

plt.legend()

plt.subplot(1,2,2)

plt.plot(hist.history['acc'],label = 'train acc')

plt.plot(hist.history['val_acc'], label = 'val acc')

plt.legend()

이러한 형식으로 네트워크를 만들 수 있다. 이건 확실하게 오버피팅이 발생한 그래프이다.