2、 Training fashion_ MNIST dataset

One 、 load fashion_mnist Data sets

fashion_mnist The data in the dataset is 28*28 The size of 10 Classified clothing dataset
One of the training sets 60000 Zhang , Test set 10000 Zhang

from tensorflow import keras
import tensorflow as tf
import matplotlib.pyplot as plt
import numpy as np

fashion_mnist = keras.datasets.fashion_mnist
(train_images,train_labels),(test_images,test_labels) = fashion_mnist.load_data()

print(train_images.shape)
""" (60000, 28, 28) """
print(test_images.shape)
""" (10000, 28, 28) """
print(train_labels.shape)
""" (60000,) """
print(test_labels.shape)
""" (60000,) """

Just look at the pixel value to see if you can guess what this picture is ？

print(train_images[0])# Look at the first picture of the training set 28*28 Pixel value 
""" [[ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 13 73 0 0 1 4 0 0 0 0 1 1 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 3 0 36 136 127 62 54 0 0 0 1 3 4 0 0 3] [ 0 0 0 0 0 0 0 0 0 0 0 0 6 0 102 204 176 134 144 123 23 0 0 0 0 12 10 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 155 236 207 178 107 156 161 109 64 23 77 130 72 15] [ 0 0 0 0 0 0 0 0 0 0 0 1 0 69 207 223 218 216 216 163 127 121 122 146 141 88 172 66] [ 0 0 0 0 0 0 0 0 0 1 1 1 0 200 232 232 233 229 223 223 215 213 164 127 123 196 229 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 183 225 216 223 228 235 227 224 222 224 221 223 245 173 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 193 228 218 213 198 180 212 210 211 213 223 220 243 202 0] [ 0 0 0 0 0 0 0 0 0 1 3 0 12 219 220 212 218 192 169 227 208 218 224 212 226 197 209 52] [ 0 0 0 0 0 0 0 0 0 0 6 0 99 244 222 220 218 203 198 221 215 213 222 220 245 119 167 56] [ 0 0 0 0 0 0 0 0 0 4 0 0 55 236 228 230 228 240 232 213 218 223 234 217 217 209 92 0] [ 0 0 1 4 6 7 2 0 0 0 0 0 237 226 217 223 222 219 222 221 216 223 229 215 218 255 77 0] [ 0 3 0 0 0 0 0 0 0 62 145 204 228 207 213 221 218 208 211 218 224 223 219 215 224 244 159 0] [ 0 0 0 0 18 44 82 107 189 228 220 222 217 226 200 205 211 230 224 234 176 188 250 248 233 238 215 0] [ 0 57 187 208 224 221 224 208 204 214 208 209 200 159 245 193 206 223 255 255 221 234 221 211 220 232 246 0] [ 3 202 228 224 221 211 211 214 205 205 205 220 240 80 150 255 229 221 188 154 191 210 204 209 222 228 225 0] [ 98 233 198 210 222 229 229 234 249 220 194 215 217 241 65 73 106 117 168 219 221 215 217 223 223 224 229 29] [ 75 204 212 204 193 205 211 225 216 185 197 206 198 213 240 195 227 245 239 223 218 212 209 222 220 221 230 67] [ 48 203 183 194 213 197 185 190 194 192 202 214 219 221 220 236 225 216 199 206 186 181 177 172 181 205 206 115] [ 0 122 219 193 179 171 183 196 204 210 213 207 211 210 200 196 194 191 195 191 198 192 176 156 167 177 210 92] [ 0 0 74 189 212 191 175 172 175 181 185 188 189 188 193 198 204 209 210 210 211 188 188 194 192 216 170 0] [ 2 0 0 0 66 200 222 237 239 242 246 243 244 221 220 193 191 179 182 182 181 176 166 168 99 58 0 0] [ 0 0 0 0 0 0 0 40 61 44 72 41 35 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]] """

Output the following photo

plt.imshow(train_images[0])

Two 、 Start training model

model = keras.Sequential([
    keras.layers.Flatten(input_shape=(28,28)),# The photo is completely flattened , One dimensional array form 
    keras.layers.Dense(128,activation=tf.nn.relu),#128 Neurons 
    keras.layers.Dense(10,activation=tf.nn.softmax)# Output layer 0-9, Ten altogether 
])

Look at the structure of the model
first floor 784 individual ,flatten Layer to be imported 2828 Expand the image , Line up ,2828=784

The second floor 128 individual ,128 Neurons ;100480 Parameters , First floor 784 And the second floor 128 Full Permutation ,784*128=100352, Each one has a bias Bias term ,100352+128=100480

The third level 10 individual , That is to say 10 classification ,10 Different categories , Output at that time 10 Probability values , What is big is what kind ;1290 Parameters , The second floor 128 Neurons , They are different from 10 Make a full arrangement ,128*10=1280, Each one has a bias Bias term ,1280+10=1290

model.summary()
""" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= flatten (Flatten) (None, 784) 0 _________________________________________________________________ dense (Dense) (None, 128) 100480 _________________________________________________________________ dense_1 (Dense) (None, 10) 1290 ================================================================= Total params: 101,770 Trainable params: 101,770 Non-trainable params: 0 _________________________________________________________________ """

In order to make the effect better , Normalize the image pixel values in the data set to 0-1 Between

train_images_y = train_images/255# Normalize the training image 

Training 50 Time

model.compile(optimizer="adam",loss="sparse_categorical_crossentropy",metrics=['accuracy'])# Specify optimization method and loss function 
model.fit(train_images_y,train_labels,epochs=50)# Training 

Because the normalized image of the training set is transmitted during model training
so , During model evaluation, it is also necessary to normalize the test set

test_images_y = test_images/255# The test image should also be normalized during test evaluation 
model.evaluate(test_images_y,test_labels)#evaluate Evaluation effect 
""" [0.5110174604289234, 0.8845] """

Select a few from the test set to test , It actually outputs 10 It's worth , That is, the probability of probability , The biggest is the category of forecasts

model.predict([[test_images[0]/255]])
""" array([[2.2063166e-16, 1.1835037e-17, 7.4574429e-23, 2.0577940e-22, 4.3680589e-17, 2.7080047e-08, 3.8249505e-15, 3.4797877e-06, 1.4701404e-10, 9.9999654e-01]], dtype=float32) """

The one with the largest predicted value by the screening model

print(np.argmax(model.predict([[test_images[0]/255]])))
""" 9 """

Look at the actual label of this picture

print(test_labels[0])
""" 9 """

The predicted value is the same as the actual value , It shows that the prediction is correct

Show me this picture

plt.imshow(train_images[0])