在Tensorflow卷積神經(jīng)網(wǎng)絡實例這篇博客中，我們實現(xiàn)了一個簡單的卷積神經(jīng)網(wǎng)絡，沒有復雜的Trick。接下來，我們將使用CIFAR-10數(shù)據(jù)集進行訓練。
CIFAR-10是一個經(jīng)典的數(shù)據(jù)集，包含60000張32*32的彩色圖像，其中訓練集50000張，測試集10000張。CIFAR-10如同其名字，一共標注為10類，每一類圖片6000張。
本文實現(xiàn)了進階的卷積神經(jīng)網(wǎng)絡來解決CIFAR-10分類問題，我們使用了一些新的技巧：
首先需要下載Tensorflow models Tensorflow models，以便使用其中的CIFAR-10數(shù)據(jù)的類.進入目錄models/tutorials/image/cifar10目錄，執(zhí)行以下代碼
import cifar10import cifar10_inputimport tensorflow as tfimport numpy as npimport time# 定義batch_size, 訓練輪數(shù)max_steps, 以及下載CIFAR-10數(shù)據(jù)的默認路徑max_steps = 3000batch_size = 128data_dir = 'E://tmp/cifar10_data/cifar-10-batches-bin'# 定義初始化weight的函數(shù)，定義的同時，對weight加一個L2 loss，放在集'losses'中def variable_with_weight_loss(shape, stddev, w1): var = tf.Variable(tf.truncated_normal(shape, stddev=stddev)) if w1 is not None: weight_loss = tf.multiply(tf.nn.l2_loss(var), w1, name='weight_loss') tf.add_to_collection('losses', weight_loss) return var# 使用cifar10類下載數(shù)據(jù)集，并解壓、展開到其默認位置#cifar10.maybe_download_and_extract()# 在使用cifar10_input類中的distorted_inputs函數(shù)產(chǎn)生訓練需要使用的數(shù)據(jù)。需要注意的是，返回的是已經(jīng)封裝好的tensor，# 且對數(shù)據(jù)進行了Data Augmentation（水平翻轉、隨機剪切、設置隨機亮度和對比度、對數(shù)據(jù)進行標準化）images_train, labels_train = cifar10_input.distorted_inputs(data_dir=data_dir, batch_size=batch_size)# 再使用cifar10_input.inputs函數(shù)生成測試數(shù)據(jù)，這里不需要進行太多處理images_test, labels_test = cifar10_input.inputs(eval_data=True, data_dir=data_dir, batch_size=batch_size)# 創(chuàng)建數(shù)據(jù)的placeholderimage_holder = tf.placeholder(tf.float32, [batch_size, 24, 24, 3])label_holder = tf.placeholder(tf.int32, [batch_size])# 創(chuàng)建第一個卷積層weight1 = variable_with_weight_loss(shape=[5, 5, 3, 64], stddev=5e-2, w1=0.0)kernel1 = tf.nn.conv2d(image_holder, weight1, strides=[1, 1, 1, 1], padding='SAME')bias1 = tf.Variable(tf.constant(0.0, shape=[64]))conv1 = tf.nn.relu(tf.nn.bias_add(kernel1, bias1))pool1 = tf.nn.max_pool(conv1, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1], padding='SAME')# LRN層對ReLU會比較有用，但不適合Sigmoid這種有固定邊界并且能抑制過大值的激活函數(shù)norm1 = tf.nn.lrn(pool1, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75)# 創(chuàng)建第二個卷積層weight2 = variable_with_weight_loss(shape=[5, 5, 64, 64], stddev=5e-2, w1=0.0)kernel2 = tf.nn.conv2d(norm1, weight2, strides=[1, 1, 1, 1], padding='SAME')bias2 = tf.Variable(tf.constant(0.1, shape=[64]))conv2 = tf.nn.relu(tf.nn.bias_add(kernel2, bias2))norm2 = tf.nn.lrn(conv2, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75)pool2 = tf.nn.max_pool(norm2, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1], padding='SAME')# 使用一個全連接層reshape = tf.reshape(pool2, [batch_size, -1])dim = reshape.get_shape()[1].valueweight3 = variable_with_weight_loss(shape=[dim, 384], stddev=0.04, w1=0.004)bias3 = tf.Variable(tf.constant(0.1, shape=[384]))local3 = tf.nn.relu(tf.matmul(reshape, weight3) + bias3)# 再使用一個全連接層，隱含節(jié)點數(shù)下降了一半，只有192個，其他的超參數(shù)保持不變weight4 = variable_with_weight_loss(shape=[384, 192], stddev=0.04, w1=0.004)bias4 = tf.Variable(tf.constant(0.1, shape=[192]))local4 = tf.nn.relu(tf.matmul(local3, weight4) + bias4)# 最后一層，將softmax放在了計算loss部分weight5 = variable_with_weight_loss(shape=[192, 10], stddev=1 / 192.0, w1=0.0)bias5 = tf.Variable(tf.constant(0.0, shape=[10]))logits = tf.add(tf.matmul(local4, weight5), bias5)# 定義lossdef loss(logits, labels): labels = tf.cast(labels, tf.int64) cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=logits, labels=labels, name='cross_entropy_per_example') cross_entropy_mean = tf.reduce_mean(cross_entropy, name='cross_entropy') tf.add_to_collection('losses', cross_entropy_mean) return tf.add_n(tf.get_collection('losses'), name='total_loss')# 獲取最終的lossloss = loss(logits, label_holder)# 優(yōu)化器train_op = tf.train.AdamOptimizer(1e-3).minimize(loss)# 使用tf.nn.in_top_k函數(shù)求輸出結果中top k的準確率，默認使用top 1，也就是輸出分數(shù)最高的那一類的準確率top_k_op = tf.nn.in_top_k(logits, label_holder, 1)# 使用tf.InteractiveSession創(chuàng)建默認的session，接著初始化全部模型參數(shù)sess = tf.InteractiveSession()tf.global_variables_initializer().run()# 啟動圖片數(shù)據(jù)增強線程tf.train.start_queue_runners()# 正式開始訓練for step in range(max_steps): start_time = time.time() image_batch, label_batch = sess.run([images_train, labels_train]) _, loss_value = sess.run([train_op, loss], feed_dict={image_holder: image_batch, label_holder: label_batch}) duration = time.time() - start_time if step % 10 == 0: example_per_sec = batch_size / duration sec_per_batch = float(duration) format_str = 'step %d, loss=%.2f ,%.1f examples/sec, %.3f sec/batch' print(format_str % (step, loss_value, example_per_sec, sec_per_batch))num_examples = 10000import mathnum_iter = int(math.ceil(num_examples / batch_size))true_count = 0total_sample_count = num_iter * batch_sizestep = 0while step < num_iter: image_batch, label_batch = sess.run([images_test, labels_test]) predictions = sess.run([top_k_op], feed_dict={image_holder: image_batch, label_holder: label_holder}) true_count += np.sum(predictions) step += 1precision = true_count / total_sample_countprint('precision @ 1 = %.3f'%precision)
新聞熱點
疑難解答
