双向循环神经网络(Bidirectional Recurrent Neural Networks,Bi-RNN),Schuster、Paliwal,1997年首次提出,和LSTM同年。Bi-RNN,增加RNN可利用信息。普通MLP,数据长度有限制。RNN,可以处理不固定长度时序数据,无法利用历史输入未来信息。Bi-RNN,同时使用时序数据输入历史及未来数据,时序相反两个循环神经网络连接同一输出,输出层可以同时获取历史未来信息。
Language Modeling,不适合Bi-RNN,目标是通过前文预测下一单词,不能将下文信息传给模型。分类问题,手写文字识别、机器翻译、蛋白结构预测,Bi-RNN提升模型效果。百度语音识别,通过Bi-RNN综合上下文语境,提升模型准确率。
Bi-RNN网络结构核心,普通单向RNN拆成两个方向,随时序正向,逆时序反赂。当前时间节点输出,同时利用正向、反向两个方向信息。两个不同方向RNN不共用state,正向RNN输出state只传给正向RNN,反向RNN输出state只传给反向RNN,正反向RNN没有直接连接。每个时间节点输入,分别传给正反向RNN,根据各自状态产生输出,两份输出一起连接到Bi-RNN输出节点,共同合成最终输出。对当前时间节点输出贡献(或loss),在训练中计算出来,参数根据梯度优化到合适值。
Bi-RNN训练,正反向RNN没有交集,分别展开普通前馈网络。BPTT(back-propagation through time)算法训练,无法同时更新状态、输出。正向state在t=1时未知,反向state在t=T时未知,state在正反向开始处未知,需人工设置。正向状态导数在t=T时未知,反向状态导数在t=1时未知,state导数在正反向结晶尾处未知,需设0代表参数更新不重要。
开始训练,第一步,输入数据forward pass操作,inference操作,先沿1->T方向计算正向RNN state,再沿T->1方向计算反向RNN state,获得输出output。第二步,backward pass操作,目标函数求导操作,先求导输出output,先沿T->1方向计算正向RNN state导数,再沿1->T方向计算反向RNN state导数。第三步,根据求得梯度值更新模型参数,完成训练。
Bi-RNN每个RNN单元,可以是传统RNN,可以是LSTM或GRU单元。可以在一层Bi-RNN上再叠加一层Bi-RNN,上层Bi-RNN输出作下层Bi-RNN输入,可以进一步抽象提炼特征。分类任务,Bi-RNN输出序列连接全连接层,或连接全局平均池化Global Average Pooling,再接Softmax层,和卷积网络一样。
TensorFlow实现Bidirectional LSTM Classifier,在MNIST数据集测试。载入TensorFlow、NumPy、TensorFlow自带MNIST数据读取器。input_data.read_data_sets下载读取MNIST数据集。
设置训练参数。设置学习速率 0.01,优化器选择Adam,学习速率低。最大训练样本数 40万,batch_size 128,设置每间隔10次训练展示训练情况。
MNIST图像尺寸 28x28,输入n_input 28(图像宽),n_steps LSTM展开步数(unrolled steps of LSTM),设28(图像高),图像全部信息用上。一次读取一行像素(28个像素点),下个时间点再传入下一行像素点。n_hidden(LSTM隐藏节点数)设256,n_classes(MNIST数据集分类数目)设10。
创建输入x和学习目标y 的place_holder。输入x每个样本直接用二维结构。样本为一个时间序列,第一维度 时间点n_steps,第二维度 每个时间点数据n_input。设置Softmax层weights和biases,tf.random_normal初始化参数。双向LSTM,forward、backward两个LSTM cell,weights参数数量翻倍,2*n_hidden。
定义Bidirectional LSTM网络生成函数。形状(batch_size,n_steps,n_input)输入变长度n_steps列表,元素形状(batch_size,n_input)。输入转置,tf.transpose(x,[1,0,2]),第一维度batch_size,第二维度n_steps,交换。tf.reshape,输入x变(n_steps*batch_size,n_input)形状。 tf.split,x拆成长度n_steps列表,列表每个tensor尺寸(batch_size,n_input),符合LSTM单元输入格式。tf.contrib.rnn.BasicLSTMCell,创建forward、backward LSTM单元,隐藏节点数设n_hidden,forget_bias设1。正向lstm_fw_cell和反向lstm_bw_cell传入Bi-RNN接口tf.nn.bidirectional_rnn,生成双向LSTM,传入x输入。双向LSTM输出结果output做矩阵乘法加偏置,参数为前面定义weights、biases。
最后输出结果,tf.nn.softmax_cross_entropy_with_logits,Softmax处理计算损失。tf.reduce_mean计算平均cost。优化器Adam,学习速率learning_rate。tf.argmax得到模型预测类别,tf.equal判断是否预测正确。tf.reduce_mean求平均准确率。
执行训练和测试操作。执行初始化参数,定义一个训练循环,保持总训练样本数(迭代数*batch_size)小于设定值。每轮训练迭代,mnist.train.next_batch拿到一个batch数据,reshape改变形状。包含输入x和训练目标y的feed_dict传入,执行训练操作,更新模型参数。迭代数display_step整数倍,计算当前batch数据预测准确率、loss,展示。
全部训练迭代结果,训练好模型,mnist.test.images全部测试数据预测,展示准确率。
完成40万样本训练,训练集预测准确率基本是1,10000样本测试集0.983准确率。
Bidirectional LSTM Classifier,MNIST数据集表现不如卷积神经网络。Bi-RNN、双向LSTM网络,时间序列分类任务表现更好,同时利用时间序列历史和未来信息,结合上下文信息,结果综合判断。
复制代码
import tensorflow as tf
import numpy as np
# Import MINST data
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
# Parameters
learning_rate = 0.01
max_samples = 400000
batch_size = 128
display_step = 10
# Network Parameters
n_input = 28 # MNIST data input (img shape: 28*28)
n_steps = 28 # timesteps
n_hidden = 256 # hidden layer num of features
n_classes = 10 # MNIST total classes (0-9 digits)
# tf Graph input
x = tf.placeholder("float", [None, n_steps, n_input])
y = tf.placeholder("float", [None, n_classes])
# Define weights
weights = {
# Hidden layer weights => 2*n_hidden because of foward + backward cells
‘out‘: tf.Variable(tf.random_normal([2*n_hidden, n_classes]))
}
biases = {
‘out‘: tf.Variable(tf.random_normal([n_classes]))
}
def BiRNN(x, weights, biases):
# Prepare data shape to match `bidirectional_rnn` function requirements
# Current data input shape: (batch_size, n_steps, n_input)
# Required shape: ‘n_steps‘ tensors list of shape (batch_size, n_input)
# Permuting batch_size and n_steps
x = tf.transpose(x, [1, 0, 2])
# Reshape to (n_steps*batch_size, n_input)
x = tf.reshape(x, [-1, n_input])
# Split to get a list of ‘n_steps‘ tensors of shape (batch_size, n_input)
x = tf.split(x, n_steps)
# Define lstm cells with tensorflow
# Forward direction cell
lstm_fw_cell = tf.contrib.rnn.BasicLSTMCell(n_hidden, forget_bias=1.0)
# Backward direction cell
lstm_bw_cell = tf.contrib.rnn.BasicLSTMCell(n_hidden, forget_bias=1.0)
# Get lstm cell output
# try:
outputs, _, _ = tf.contrib.rnn.static_bidirectional_rnn(lstm_fw_cell, lstm_bw_cell, x,
dtype=tf.float32)
# except Exception: # Old TensorFlow version only returns outputs not states
# outputs = rnn.bidirectional_rnn(lstm_fw_cell, lstm_bw_cell, x,
# dtype=tf.float32)
# Linear activation, using rnn inner loop last output
return tf.matmul(outputs[-1], weights[‘out‘]) + biases[‘out‘]
pred = BiRNN(x, weights, biases)
# Define loss and optimizer
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred, labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)
# Evaluate model
correct_pred = tf.equal(tf.argmax(pred,1), tf.argmax(y,1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
# Initializing the variables
init = tf.global_variables_initializer()
# Launch the graph
with tf.Session() as sess:
sess.run(init)
step = 1
# Keep training until reach max iterations
while step * batch_size < max_samples:
batch_x, batch_y = mnist.train.next_batch(batch_size)
# Reshape data to get 28 seq of 28 elements
batch_x = batch_x.reshape((batch_size, n_steps, n_input))
# Run optimization op (backprop)
sess.run(optimizer, feed_dict={x: batch_x, y: batch_y})
if step % display_step == 0:
# Calculate batch accuracy
acc = sess.run(accuracy, feed_dict={x: batch_x, y: batch_y})
# Calculate batch loss
loss = sess.run(cost, feed_dict={x: batch_x, y: batch_y})
print("Iter " + str(step*batch_size) + ", Minibatch Loss= " + "{:.6f}".format(loss) + ", Training Accuracy= " + "{:.5f}".format(acc))
step += 1
print("Optimization Finished!")
# Calculate accuracy for 128 mnist test images
test_len = 10000
test_data = mnist.test.images[:test_len].reshape((-1, n_steps, n_input))
test_label = mnist.test.labels[:test_len]
print("Testing Accuracy:", sess.run(accuracy, feed_dict={x: test_data, y: test_label}))
原文地址:https://www.cnblogs.com/ldt-/p/10291506.html