回答編集履歴

3

d

2019/05/28 05:59

投稿

tiitoi
tiitoi

スコア21956

test CHANGED
@@ -105,3 +105,77 @@
105
105
  )
106
106
 
107
107
  ```
108
+
109
+
110
+
111
+ ## input_shape を指定するバージョン
112
+
113
+
114
+
115
+ ```python
116
+
117
+ from tensorflow import keras
118
+
119
+
120
+
121
+ batch_size = 128
122
+
123
+ num_class = 10
124
+
125
+ epochs = 20
126
+
127
+
128
+
129
+ (x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()
130
+
131
+
132
+
133
+ x_train, x_test = x_train / 255.0, x_test / 255.0
134
+
135
+ x_train = x_train.reshape(len(x_train), -1)
136
+
137
+ x_test = x_test.reshape(len(x_test), -1)
138
+
139
+
140
+
141
+ y_train = keras.utils.to_categorical(y_train)
142
+
143
+ y_test = keras.utils.to_categorical(y_test)
144
+
145
+
146
+
147
+ model = keras.models.Sequential(
148
+
149
+ [
150
+
151
+ keras.layers.Dense(512, activation="relu", input_shape=(784,)),
152
+
153
+ keras.layers.Dropout(0.2),
154
+
155
+ keras.layers.Dense(512, activation="relu"),
156
+
157
+ keras.layers.Dropout(0.2),
158
+
159
+ keras.layers.Dense(num_class, activation="softmax"),
160
+
161
+ ]
162
+
163
+ )
164
+
165
+ model.compile(loss="categorical_crossentropy", optimizer="adam", metrics=["accuracy"])
166
+
167
+ model.fit(
168
+
169
+ x_train,
170
+
171
+ y_train,
172
+
173
+ batch_size=batch_size,
174
+
175
+ epochs=epochs,
176
+
177
+ validation_data=(x_test, y_test),
178
+
179
+ )
180
+
181
+ ```

2

d

2019/05/28 05:59

投稿

tiitoi
tiitoi

スコア21956

test CHANGED
@@ -31,3 +31,77 @@
31
31
 
32
32
 
33
33
  [kerasのmnistのサンプルを読んでみる - Qiita](https://qiita.com/ash8h/items/29e24fc617b832fba136)
34
+
35
+
36
+
37
+ ----
38
+
39
+
40
+
41
+ ## 追記
42
+
43
+
44
+
45
+ 自分の環境 TensorFlow `2.0.0-alpha0` では以下のコードで動きました。
46
+
47
+
48
+
49
+ ```python
50
+
51
+ from tensorflow import keras
52
+
53
+
54
+
55
+ batch_size = 128
56
+
57
+ num_class = 10
58
+
59
+ epochs = 20
60
+
61
+
62
+
63
+ (x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()
64
+
65
+
66
+
67
+ x_train, x_test = x_train / 255.0, x_test / 255.0
68
+
69
+ y_train = keras.utils.to_categorical(y_train)
70
+
71
+ y_test = keras.utils.to_categorical(y_test)
72
+
73
+
74
+
75
+ model = keras.models.Sequential(
76
+
77
+ [
78
+
79
+ keras.layers.Flatten(),
80
+
81
+ keras.layers.Dense(512, activation="relu"),
82
+
83
+ keras.layers.Dropout(0.2),
84
+
85
+ keras.layers.Dense(10, activation="softmax"),
86
+
87
+ ]
88
+
89
+ )
90
+
91
+ model.compile(loss="categorical_crossentropy", optimizer="adam", metrics=["accuracy"])
92
+
93
+ model.fit(
94
+
95
+ x_train,
96
+
97
+ y_train,
98
+
99
+ batch_size=256,
100
+
101
+ epochs=10,
102
+
103
+ validation_data=(x_test, y_test),
104
+
105
+ )
106
+
107
+ ```

1

d

2019/05/28 05:16

投稿

tiitoi
tiitoi

スコア21956

test CHANGED
@@ -30,4 +30,4 @@
30
30
 
31
31
 
32
32
 
33
- [Python - 数字の画像を学習させるプログラムで把握できないエラーが出ます;|teratail](https://teratail.com/questions/191745)
33
+ [kerasのmnistのサンプルを読んでみる - Qiita](https://qiita.com/ash8h/items/29e24fc617b832fba136)