-
Notifications
You must be signed in to change notification settings - Fork 1
/
3-Create_Model-Tensorflow_2.py
166 lines (140 loc) · 6.43 KB
/
3-Create_Model-Tensorflow_2.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
"""
Created on Mon Oct 01 22:10:52 2018
Updated on Sun Aug 14 18:22:12 2022
Authors: Ben Wolfaardt
Inspiration for architecture taken from:
- https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/guide/tpu.ipynb#scrollTo=mgUC6A-zCMEr
- https://www.tensorflow.org/tutorials/distribute/custom_training
Multiple GPUs, Machines, TPUs implementation:
- https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/distribute/custom_training.ipynb
- https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/guide/tpu.ipynb#scrollTo=mgUC6A-zCMEr
Using Tensorboard
- https://www.tensorflow.org/tensorboard/get_started
"""
import numpy as np
import pickle
import tensorflow as tf
import datetime
from tensorflow.python.keras import Model, layers
EXPERIMENT = "Siobhan"
EPOCHS = 50
DIRECTORY_PICKLE_DATA_INPUT = "/Users/james.wolfaardt/code/__ben/Code/Deep_Learning-EEG_Data/outputs/pickles"
participant = [1]
# Load preprocessed training/test pickles
def load_data():
with open(f"{DIRECTORY_PICKLE_DATA_INPUT}/X-{participant}-Training.pickle", 'rb') as f:
X = pickle.load(f) # Shape: (1369, 63, 450, 1)
with open(f"{DIRECTORY_PICKLE_DATA_INPUT}/y-{participant}-Training.pickle", 'rb') as f:
y = pickle.load(f)
y = np.transpose(y) # Shape: (1369,)
with open(f"{DIRECTORY_PICKLE_DATA_INPUT}/X-{participant}-Test.pickle", 'rb') as f:
X_val = pickle.load(f) # Shape: (158, 63, 450, 1)
with open(f"{DIRECTORY_PICKLE_DATA_INPUT}/y-{participant}-Test.pickle", 'rb') as f:
y_val = pickle.load(f)
y_val = np.transpose(y_val) # Shape: (158,)
train_ds = tf.data.Dataset.from_tensor_slices((X, y)).shuffle(10).batch(64)
val_ds = tf.data.Dataset.from_tensor_slices((X_val, y_val)).batch(64)
return train_ds, val_ds
class MyModel(Model):
def __init__(self):
super(MyModel, self).__init__()
# https://www.tensorflow.org/api_docs/python/tf/keras/layers/Conv2D
self.conv1 = layers.Conv2D(
filters=128,
kernel_size=(3, 3),
input_shape=(128, 257, 1),
padding='same',
activation=tf.nn.relu
)
self.conv2 = layers.Conv2D(
filters=128, kernel_size=(3, 3), padding='same', activation=tf.nn.relu
)
self.conv3 = layers.Conv2D(
filters=32, kernel_size=(3, 3), padding='same', activation=tf.nn.relu
)
self.conv4 = layers.Conv2D(
filters=16, kernel_size=(3, 3), padding='same', activation=tf.nn.relu
)
# https://www.tensorflow.org/api_docs/python/tf/keras/layers/Dense
self.dense1 = layers.Dense(16, activation=tf.nn.relu)
self.dense2 = layers.Dense(8, activation=tf.nn.relu)
self.dense3 = layers.Dense(2, activation=tf.nn.sigmoid)
# https://www.tensorflow.org/api_docs/python/tf/keras/layers/Flatten
self.flatten = layers.Flatten()
# https://www.tensorflow.org/api_docs/python/tf/keras/layers/MaxPool2D
self.maxpooling = layers.MaxPool2D(pool_size=(2, 2), padding='same')
# https://www.tensorflow.org/api_docs/python/tf/keras/layers/Dropout
self.dropout = layers.Dropout(0.2)
def call(self, x):
x = self.conv1(x)
x = self.maxpooling(x)
x = self.conv2(x)
x = self.conv3(x)
x = self.maxpooling(x)
x = self.dropout(x)
x = self.conv4(x)
x = self.maxpooling(x)
x = self.dropout(x)
x = self.flatten(x)
x = self.dense1(x)
x = self.dense2(x)
x = self.dropout(x)
return self.dense3(x)
def main():
training_ds, validation_ds = load_data()
model = MyModel()
# Selected loss and optimizer:
loss_object = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False)
optimizer = tf.keras.optimizers.Adam()
# Define our metrics
training_loss = tf.keras.metrics.Mean(name='training_loss', dtype=tf.float32)
training_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(name='train_accuracy')
validation_loss = tf.keras.metrics.Mean(name='test_loss', dtype=tf.float32)
validation_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(name='test_accuracy')
@tf.function
def training_step(model, optimizer, x_train, y_train): # where x = epochs, and y = labes
with tf.GradientTape() as tape:
predictions = model(x_train, training=True)
loss = loss_object(y_train, predictions)
gradients = tape.gradient(loss, model.trainable_variables)
optimizer.apply_gradients(zip(gradients, model.trainable_variables))
training_loss(loss)
training_accuracy(y_train, predictions)
@tf.function
def validation_step(model, x_test, y_test):
predictions = model(x_test)
loss = loss_object(y_test, predictions)
validation_loss(loss)
validation_accuracy(y_test, predictions)
current_time = datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
training_log_dir = 'logs/gradient_tape/' + current_time + '/training'
test_log_dir = 'logs/gradient_tape/' + current_time + '/test'
train_summary_writer = tf.summary.create_file_writer(training_log_dir)
test_summary_writer = tf.summary.create_file_writer(test_log_dir)
for epoch in range(EPOCHS):
for (x_train, y_train) in training_ds:
training_step(model, optimizer, x_train, y_train)
with train_summary_writer.as_default():
tf.summary.scalar('loss', training_loss.result(), step=epoch)
tf.summary.scalar('accuracy', training_accuracy.result(), step=epoch)
for (x_val, y_val) in validation_ds:
validation_step(model, x_val, y_val)
with test_summary_writer.as_default():
tf.summary.scalar('loss', validation_loss.result(), step=epoch)
tf.summary.scalar('accuracy', validation_accuracy.result(), step=epoch)
template = 'Epoch {}, Loss: {}, Accuracy: {}, Test Loss: {}, Test Accuracy: {}'
print(template.format(
epoch + 1,
training_loss.result(),
training_accuracy.result() * 100,
validation_loss.result(),
validation_accuracy.result() * 100
))
# Reset metrics every epcoh
training_loss.reset_states()
training_accuracy.reset_states()
validation_loss.reset_states()
validation_accuracy.reset_states()
if __name__ == '__main__':
main()
print("Done")