Python DNN 多分类
时间: 2023-12-05 17:38:59 浏览: 41
以下是Python DNN多分类的示例代码:
```python
import tensorflow as tf
import numpy as np
import time
# 准备数据
X_train = np.random.rand(1000, 10)
train_label = np.random.randint(0, 5, size=(1000, 1))
X_test = np.random.rand(200, 10)
test_label = np.random.randint(0, 5, size=(200, 1))
# 定义网络结构
n_input = 10
n_hidden_1 = 20
n_hidden_2 = 20
n_classes = 5
x_tf = tf.placeholder(tf.float32, [None, n_input])
y_tf = tf.placeholder(tf.int32, [None, 1])
weights = {
'h1': tf.Variable(tf.random_normal([n_input, n_hidden_1])),
'h2': tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2])),
'out': tf.Variable(tf.random_normal([n_hidden_2, n_classes]))
}
biases = {
'b1': tf.Variable(tf.random_normal([n_hidden_1])),
'b2': tf.Variable(tf.random_normal([n_hidden_2])),
'out': tf.Variable(tf.random_normal([n_classes]))
}
# 定义模型
def multilayer_perceptron(x):
layer_1 = tf.add(tf.matmul(x, weights['h1']), biases['b1'])
layer_1 = tf.nn.relu(layer_1)
layer_2 = tf.add(tf.matmul(layer_1, weights['h2']), biases['b2'])
layer_2 = tf.nn.relu(layer_2)
out_layer = tf.matmul(layer_2, weights['out']) + biases['out']
return out_layer
# 定义损失函数和优化器
logits = multilayer_perceptron(x_tf)
loss_op = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(logits=logits, labels=tf.squeeze(y_tf)))
optimizer = tf.train.AdamOptimizer(learning_rate=0.01)
train_op = optimizer.minimize(loss_op)
# 定义评估模型的指标
correct_pred = tf.equal(tf.argmax(logits, 1), tf.squeeze(y_tf))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
# 训练模型
sess = tf.Session()
init = tf.global_variables_initializer()
sess.run(init)
tf_dict = {x_tf:X_train, y_tf:train_label}
start_time = time.time()
nIter = 1000
for it in range(nIter):
sess.run(train_op, tf_dict)
if it % 100 == 0:
elapsed = time.time() - start_time
loss_value = sess.run(loss_op, tf_dict)
print('It: %d, Loss: %.7e, Time: %.2f' % (it, loss_value, elapsed))
start_time = time.time()
# 测试模型
accuracy = sess.run(accuracy, feed_dict={x_tf:X_test,y_tf:test_label})
print("Accuracy: %.2f" % accuracy)
```