-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathA1-T1-cnn.py
195 lines (143 loc) · 6.69 KB
/
A1-T1-cnn.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
#!/usr/bin/env python
# coding: utf-8
# In[1]:
import keras
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import tensorflow as tf
fashion_mnist = keras.datasets.fashion_mnist
cifar10=keras.datasets.cifar10
(X_train_full, y_train_full), (X_test, y_test) = cifar10.load_data()
#create the validation set
X_valid, X_train = X_train_full[:6000] / 255.0, X_train_full[6000:] / 255.0
y_valid, y_train = y_train_full[:6000], y_train_full[6000:]
#add label for class
class_names = ["airplane","automobile","bird","cat","deer","dog","frog","horse","ship","truck"
]
# In[2]:
kinit='glorot_normal'
lr=0.001
dr=0.1
#a convolutional neural network described in Ch. 14, p. 447.
model = keras.models.Sequential([
keras.layers.Conv2D(64, 7, activation="relu",kernel_initializer=kinit, padding="same",
input_shape=[32, 32, 3]),
keras.layers.MaxPooling2D(2),
keras.layers.Conv2D(128, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.Conv2D(128, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.MaxPooling2D(2),
keras.layers.Conv2D(256, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.Conv2D(256, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.MaxPooling2D(2),
keras.layers.Flatten(),
keras.layers.Dense(128, activation="relu",kernel_initializer=kinit),
keras.layers.Dropout(dr),
keras.layers.Dense(64, activation="relu",kernel_initializer=kinit),
keras.layers.Dropout(dr),
keras.layers.Dense(10, activation="softmax")
])
opt=keras.optimizers.Adam(learning_rate=lr)
model.compile(loss="sparse_categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
history = model.fit(X_train, y_train, epochs=30,
validation_data=(X_valid, y_valid))
# In[3]:
pd.DataFrame(history.history).plot(figsize=(8, 5))
plt.grid(True)
plt.gca().set_ylim(0, 1) # set the vertical range to [0-1] plt.show()
# In[4]:
model.evaluate(X_test, y_test)
X_new = X_test[:3]
y_proba = model.predict(X_new)# get prediction probabilities
print(y_proba.round(2))
y_pred = np.argmax(y_proba, axis=-1) # get predicted class index by finding the index of maximum probability along the last axis
print(np.array(class_names)[y_pred])
print(np.array(class_names)[y_test[:3]])
#The predict_classes method was deprecated in TensorFlow 2.6 and has been removed in later versions.
#Instead, you should use model.predict followed by np.argmax to get the predicted class indices.
# In[5]:
kinit='glorot_uniform'
lr=0.001
dr=0.1
#a convolutional neural network described in Ch. 14, p. 447.
model = keras.models.Sequential([
keras.layers.Conv2D(64, 7, activation="relu",kernel_initializer=kinit, padding="same",
input_shape=[32, 32, 3]),
keras.layers.MaxPooling2D(2),
keras.layers.Conv2D(128, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.Conv2D(128, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.MaxPooling2D(2),
keras.layers.Conv2D(256, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.Conv2D(256, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.MaxPooling2D(2),
keras.layers.Flatten(),
keras.layers.Dense(128, activation="relu",kernel_initializer=kinit),
keras.layers.Dropout(dr),
keras.layers.Dense(64, activation="relu",kernel_initializer=kinit),
keras.layers.Dropout(dr),
keras.layers.Dense(10, activation="softmax")
])
opt=keras.optimizers.Adam(learning_rate=lr)
model.compile(loss="sparse_categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
history = model.fit(X_train, y_train, epochs=30,
validation_data=(X_valid, y_valid))
# In[6]:
pd.DataFrame(history.history).plot(figsize=(8, 5))
plt.grid(True)
plt.gca().set_ylim(0, 1) # set the vertical range to [0-1] plt.show()
# In[7]:
model.evaluate(X_test, y_test)
X_new = X_test[:3]
y_proba = model.predict(X_new)# get prediction probabilities
print(y_proba.round(2))
y_pred = np.argmax(y_proba, axis=-1) # get predicted class index by finding the index of maximum probability along the last axis
print(np.array(class_names)[y_pred])
print(np.array(class_names)[y_test[:3]])
#The predict_classes method was deprecated in TensorFlow 2.6 and has been removed in later versions.
#Instead, you should use model.predict followed by np.argmax to get the predicted class indices.
# In[8]:
kinit='he_uniform'
lr=0.001
dr=0.1
#a convolutional neural network described in Ch. 14, p. 447.
model = keras.models.Sequential([
keras.layers.Conv2D(64, 7, activation="relu",kernel_initializer=kinit, padding="same",
input_shape=[32, 32,3]),
keras.layers.MaxPooling2D(2),
keras.layers.Conv2D(128, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.Conv2D(128, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.MaxPooling2D(2),
keras.layers.Conv2D(256, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.Conv2D(256, 3, activation="relu",kernel_initializer=kinit, padding="same"),
keras.layers.MaxPooling2D(2),
keras.layers.Flatten(),
keras.layers.Dense(128, activation="relu",kernel_initializer=kinit),
keras.layers.Dropout(dr),
keras.layers.Dense(64, activation="relu",kernel_initializer=kinit),
keras.layers.Dropout(dr),
keras.layers.Dense(10, activation="softmax")
])
opt=keras.optimizers.Adam(learning_rate=lr)
model.compile(loss="sparse_categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
history = model.fit(X_train, y_train, epochs=30,
validation_data=(X_valid, y_valid))
# In[9]:
pd.DataFrame(history.history).plot(figsize=(8, 5))
plt.grid(True)
plt.gca().set_ylim(0, 1) # set the vertical range to [0-1] plt.show()
# In[10]:
model.evaluate(X_test, y_test)
X_new = X_test[:3]
y_proba = model.predict(X_new)# get prediction probabilities
print(y_proba.round(2))
y_pred = np.argmax(y_proba, axis=-1) # get predicted class index by finding the index of maximum probability along the last axis
print(np.array(class_names)[y_pred])
print(np.array(class_names)[y_test[:3]])
#The predict_classes method was deprecated in TensorFlow 2.6 and has been removed in later versions.
#Instead, you should use model.predict followed by np.argmax to get the predicted class indices.