应用TensorFlow高级api构建全连接神经网络(1)

应用TensorFlow高级api构建全连接神经网络(1)

应用TensorFlow高级api构建全连接神经网络

.ipynb

import numpy as np
import tensorflow as ist import input_data
mnist = ad_data_sets("/tmp/data/", one_hot=False)# 参数
learning_rate = 0.1
num_steps = 1000
batch_size = 128
display_step = 100# 网络参数
num_input = 784
n_hidden_1 = 256
n_hidden_2 = 256
num_classes = 10# 定义训练的输入函数
input_fn = tf.estimator.inputs.numpy_input_fn(x={'images': ain.images}, y&#ain.labels,batch_size=batch_size, num_epochs=None, shuffle=True)# 定义神经网络
def neural_net(x_dict):x = x_dict['images']#kernel_initializer：可选，默认为 None，即权重的初始化方法，如果为 None，则使用默认的 Xavier 初始化方法。layer_1 = tf.layers.dense(x, n_hidden_1)layer_2 = tf.layers.dense(layer_1, n_hidden_2)out_layer = tf.layers.dense(layer_2, num_classes)return out_layer# 定义模型函数
def model_fn(features, labels, mode):logits = neural_net(features)# 返回的是logits中的最大值的索引号pred_classes = tf.argmax(logits, axis=1)pred_probas = tf.nn.softmax(logits)# 预测模式if mode == tf.estimator.ModeKeys.PREDICT:return tf.estimator.EstimatorSpec(mode, predictions=pred_classes)# 训练模式# 损失函数loss_op = tf.reduce_sparse_softmax_cross_entropy_with_logits(logits=logits, label=tf.cast(labels, dtype=tf.int32)))# 优化器optimizer = tf.train.GradientDescentOptimizer(learning_rate = learning_rate) train_op = optimizer.minimize(loss_op, global_step&#_global_step())# 计算准确率acc_op = tf.metrics.accuracy(labels=labels, predictions=pred_classes)estim_specs = tf.estimator.EstimatorSpec(mode=mode,predictions=pred_classes,loss=loss_op,train_op=train_op,eval_metric_ops={'accuracy':acc_op})return estim_specs# 构建评价器
model = tf.estimator.Estimator(model_fn)
# 训练模型
ain(input_fn, steps=num_steps)
# 评价模型
input_fn = tf.estimator.inputs.numpy_input_fn(x={'images': st.images}, y&#st.labels,batch_size=batch_size, shuffle=False)
model.evaluate(input_fn)# 预测
n_images = 4
test_images = st.images[:n_images]
input_fn = tf.estimator.inputs.numpy_input_fn(x={'images': test_images}, shuffle=False)
preds = list(model.predict(input_fn))for i in range(n_images):print("Model prediction:", preds[i])

高级api解释见下节。