TensorFlow 实战卷积神经网络之 LeNet

欢迎大家关注我们的网站和系列教程：http://www.tensorflownews.com/，学习更多的机器学习、深度学习的知识！

LeNet

项目简介

1994 年深度学习三巨头之一的 Yan LeCun 提出了 LeNet 神经网络，这是最早的卷积神经网络。1998 年 Yan LeCun 在论文 “Gradient- d Learning Applied to Document Recognition” 中将这种卷积神经网络命名为 “LeNet-5”。LeNet 已经包含了现在卷积神经网络中的卷积层，池化层，全连接层，已经具备了卷积神经网络必须的基本组件。

Gradient- d Learning Applied to Document Recognition
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=726791

Architecture of LeNet-5 (Convolutional Neural Networks) for digit recognition

数据处理

同卷积神经网络中的 MNIST 数据集处理方法。

TensorFlow 卷积神经网络手写数字识别数据集介绍

http://www.tensorflownews.com/2018/03/26/tensorflow-mnist/

模型实现

经典的卷积神经网络，TensorFlow 官方已经实现，并且封装在了 tensorflow 库中，以下内容截取自 TensorFlow 官方 Github。

models/research/slim/nets/lenet.py
https://github.com/tensorflow/models/blob/master/research/slim/nets/lenet.py

import tensorflow as tfslim = tf.contrib.slimdef lenet(images, num_classes=10, is_training=False,          dropout_keep_prob=0.5,          prediction_fn=slim.softmax,          scope='LeNet'):  end_points = {}  with tf.variable_scope(scope, 'LeNet', [images]):    net = end_points['conv1'] = slim.conv2d(images, 32, [5, 5], scope='conv1')    net = end_points['pool1'] = slim.max_pool2d(net, [2, 2], 2, scope='pool1')    net = end_points['conv2'] = slim.conv2d(net, 64, [5, 5], scope='conv2')    net = end_points['pool2'] = slim.max_pool2d(net, [2, 2], 2, scope='pool2')    net = slim.flatten(net)    end_points['Flatten'] = net    net = end_points['fc3'] = slim.fully_connected(net, 1024, scope='fc3')    if not num_classes:      return net, end_points    net = end_points['dropout3'] = slim.dropout(        net, dropout_keep_prob, is_training=is_training, scope='dropout3')    logits = end_points['Logits'] = slim.fully_connected(        net, num_classes, activation_fn=None, scope='fc4')  end_points['Predictions'] = prediction_fn(logits, scope='Predictions')  return logits, end_pointslenet.default_image_size = 28def lenet_arg_scope(weight_decay=0.0):  """Defines the default lenet argument scope.  Args:    weight_decay: The weight decay to use for regularizing the model.  Returns:    An `arg_scope` to use for the inception v3 model.  """  with slim.arg_scope(      [slim.conv2d, slim.fully_connected],      weights_regularizer=slim.l2_regularizer(weight_decay),      weights_initializer=tf.truncated_normal_initializer(stddev=0.1),      activation_fn=tf.nn.relu) as sc:    return sc

模型优化

欢迎大家关注我们的网站和系列教程：http://www.tensorflownews.com/，学习更多的机器学习、深度学习的知识！