模型图构建流程
导包
1
2
3
4
5
6
7
| import tensorflow as tf
import os
import pickle
import numpy as np
CIFAR_DIR = 'cifar-10-batches-py'
print(os.listdir(CIFAR_DIR))
|
[‘batches.meta’, ‘test_batch’, ‘data_batch_1’, ‘data_batch_4’, ‘data_batch_5’, ‘readme.html’, ‘data_batch_2’, ‘data_batch_3’]
读数据模块
1
2
3
4
5
| def load_data(filename):
"""read date from data file."""
with open(filename,'rb') as f:
data = pickle.load(f)
return data['data'],data['labels']
|
模型图构建
神经元
1
2
3
4
5
6
7
8
9
10
11
12
| # 输入 [None,3072]
x = tf.placeholder(tf.float32,[None,3072])
# 输出 [None]
y = f.placeholder(tf.int64,[None])
# 权重 (3072,1)
w = tf.get_variable('w',[x.get_shape()[-1],1],
initializer=tf.random_normal_initializer(0,1))
# 偏置 (1, )
b = tf.get_variable('b',[1],initializer=tf.constant_initializer(0.0))
|
激活函数
1
2
| # [None,3072]*[3072,1] = [None,1]
y_ =tf.matmul(x,w)+b
|
逻辑回归模型
1
2
| # [None,1]
p_y_1 = tf.nn.sigmoid(y_)
|
损失函数
1
2
3
4
5
6
7
8
9
| # [None,1]
y_reshaped = tf.reshape(y,(-1,1))
y_reshaped_float = tf.cast(y_reshaped,tf.float32)
loss = tf.reduce_mean(tf.square(y_reshaped - p_y_1))
#bool
predict = p_y_1 > 0.5
# [1,0,1,1,1,0]
correct_prediction = tf.equal(tf.cast(predict,tf.int64),y_reshaped)
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float64))
|
神经网络训练
1
2
| with tf.name_scope('train_op'):
train_op = tf.train.AdamOptimizer(1e-3).minimize(loss)
|
