[Tensorflow] RNN - 04. Work with CNN for Text Classification
Ref: Combining CNN and RNN for spoken language identification
Ref: Convolutional Methods for Text
[1] CONVOLUTIONAL, LONG SHORT-TERM MEMORY, FULLY CONNECTED DEEP NEURAL NETWORKS
[2] Efficient Character-level Document Classification by Combining Convolution and Recurrent Layers
结合此文,对sentiment prediction做进一步的性能提升。
一个近似的例子:https://github.com/LopezGG/NN_NER_tensorFlow/blob/master/network.py
#Embedding layer (is always built on CPU. There is bug that makes embedding fail on GPU)
with tf.device('/cpu:0'), tf.name_scope("char_embedding"):
#plus 1 becuase 0 is for unknown char
self.W_char = tf.Variable(tf.random_uniform([char_vocab_size+1, char_embedd_dim],-1,1),trainable=True, name="W_char")
self.char_embedding_placeholder = tf.placeholder(tf.float32, [char_vocab_size+1, char_embedd_dim])
char_embedding_init = self.W_char.assign(self.char_embedding_placeholder)
self.embedded_char = tf.nn.embedding_lookup(self.W_char, self.input_x_char_flat,name="embedded_char") #shape [batch_size,max_char_per_word*sequence_length,char_embedd_dim]
self.embedded_char_dropout =tf.nn.dropout(self.embedded_char, self.dropout_keep_prob,name="embedded_char_dropout")
#Add CNN get filters and combine with word
with tf.name_scope("char_conv_maxPool"):
filter_shape = [filter_size, char_embedd_dim, num_filters]
W_conv = tf.Variable(tf.truncated_normal(filter_shape, stddev=0.1), name="W_conv")
b_conv = tf.Variable(tf.constant(0.1, shape=[num_filters]), name="b_conv") conv = tf.nn.conv1d(self.embedded_char_dropout,
W_conv,
stride=1,
padding="SAME",
name="conv") #will have dimensions [batch_size,out_width,num_filters] out_width is a function of max_words,filter_size and stride_size #(?, 3051, 20)
#out_width for same padding iwth stride 1 given by (max_char_per_word*sequence_length)
print("conv.get_Shape(): ",conv.get_shape())
# Apply nonlinearity TODO: Test without relu
#h = tf.nn.bias_add(conv, b_conv,name="add bias")#does not change dimensions
h_expand = tf.expand_dims(conv, -1)
print("h_expand.get_Shape(): ",h_expand.get_shape())
pooled = tf.nn.max_pool(
h_expand,
#[batch, height, width, channels]
ksize=[1,sequence_length * max_char_per_word,1, 1], #On the batch size dimension and the channels dimension, ksize is 1 because we don't want to take the maximum over multiple examples, or over multiples channels.
strides=[1, max_char_per_word, 1, 1],
padding='SAME',
name="pooled")
#print("pooled.get_Shape(): ",pooled.get_shape())
#[batch_size,(max_char_per_word*sequence_length), num_filters, 1] --> [batch, sequence_length, num_filters] , same as word_embedding layer (?, 113, 20, 1) --> (?, 113, 20)
self.char_pool_flat = tf.reshape(pooled, [-1,sequence_length,num_filters],name="char_pool_flat")
#print("self.char_pool_flat.get_shape(): ",self.char_pool_flat.get_shape())
#[batch, sequence_length, word_embedd_dim+num_filters]
self.word_char_features = tf.concat([self.embedded_words, self.char_pool_flat], axis=2) #we mean that the feature with index 2 i/e num_filters is variable
#print("self.word_char_features.get_shape(): ",self.word_char_features.get_shape())
self.word_char_features_dropout =tf.nn.dropout(self.word_char_features, self.dropout_keep_prob,name="word_char_features_dropout")
Combinations of CNN and RNN
The general architecture of these combinations is a convolutional feature extractor applied on the input, then some recurrent network on top of the CNN’s output, then an optional fully connected layer on RNN’s output and finally a softmax layer.
The output of the CNN is a set of several channels (also known as feature maps). We can have separate GRUs acting on each channel (with or without weight sharing) as described in this picture (Left).
Another option is to interpret CNN’s output as a 3D-tensor and run a single GRU on 2D slices of that tensor, picture (Right).
The latter option has more parameters,but the information from different channels is mixed inside the GRU, and it seems to improve performance. 【后者貌似好】
This architecture is similar to the one described in this paper[1] on speech recognition,
except that they also use some residual connections (“shortcuts”) from input to RNN and from CNN to fully connected layers.
It is interesting to note that recently it was shown that similar architectures work well for text classification[2].
Network | Accuracy | Notes |
---|---|---|
tc_net_rnn | 92.4 | CNN consists of 3 convolutional blocks and outputs 32 channels of size 104x13. Each of these channels is fed to a separate GRU as a sequence of 104 vectors of size 13. The outputs of GRUs are combined and fed to a fully connected layer |
tc_net_rnn_nodense | 91.94 | Same as above, except there is no fully connected layer on top of GRUs. Outputs of GRU are fed directly to the softmax layer |
tc_net_rnn_shared | 96.96 | Same as above, but the 32 GRUs share weights. This helped to fight overfitting |
tc_net_rnn_shared_pad | 98.11 | 4 convolutional blocks in CNN using pad=2 instead of ignore_broder=False (which enabled CuDNN and the training became much faster). The output of CNN is a set of 32 channels of size 54x8. 32 GRUs are applied (one for each channel) with shared weights and there is no fully connected layer |
tc_net_deeprnn_shared_pad | 95.67 | 4 convolutional block as above, but 2-layer GRUs with shared weights are applied on CNN’s outputs. Overfitting became stronger because of this second layer |
tc_net_shared_pad_augm | 98.68 | Same as tc_net_rnn_shared_pad, but the network randomly crops the input and takes 9s interval. The performance became a bit better due to this |
tc_net_rnn_onernn | 99.2 | The outputs of a CNN with 4 convolutional blocks are grouped into a 32x54x8 3D-tensor and a single GRU runs on a sequence of 54 vectors of size 32*8 |
tc_net_rnn_onernn_notimepool | 99.24 | Same as above, but the stride along the time axis is set to 1 in every pooling layer. Because of this the CNN outputs 32 channels of size 852x8 |
The second layer of GRU in this setup didn’t help due to the overfitting.
It seems that subsampling in the time dimension is not a good idea. The information that is lost during subsampling can be better used by the RNN.
In the paper on text classification[2] by Yijun Xiao and Kyunghyun Cho, the authors even suggest that maybe all pooling/subsampling layers can be replaced by recurrent layers.
训练技巧
These networks were trained using SGD with momentum only. The learning rate was set to 0.003 for around 10 epochs, then it was manually decreased to 0.001 and then to 0.0003. On average, it took 35 epochs to train these networks.
Ensembling
The best single model had 99.24% accuracy on the validation set. We had 33 predictions by all these models (there were more than one predictions for some models, taken after different epochs) and we just summed up the predicted probabilities and got 99.67% accuracy. Surprisingly, our other attempts of ensembling (e.g. majority voting, ensemble only on some subset of all models) didn’t give better results.
[Tensorflow] RNN - 04. Work with CNN for Text Classification的更多相关文章
- [转] Implementing a CNN for Text Classification in TensorFlow
Github上的一个开源项目,文档讲得极清晰 Github - https://github.com/dennybritz/cnn-text-classification-tf 原文- http:// ...
- Implementing a CNN for Text Classification in TensorFlow
参考: 1.Understanding Convolutional Neural Networks for NLP 2.Implementing a CNN for Text Classificati ...
- CNN tensorflow text classification CNN文本分类的例子
from:http://deeplearning.lipingyang.org/tensorflow-examples-text/ TensorFlow examples (text-based) T ...
- [Tensorflow] RNN - 01. Spam Prediction with BasicRNNCell
Ref: http://blog.csdn.net/mebiuw/article/details/60780813 Ref: https://medium.com/@erikhallstrm/hell ...
- TensorFlow (RNN)深度学习 双向LSTM(BiLSTM)+CRF 实现 sequence labeling 序列标注问题 源码下载
http://blog.csdn.net/scotfield_msn/article/details/60339415 在TensorFlow (RNN)深度学习下 双向LSTM(BiLSTM)+CR ...
- tensorflow rnn 最简单实现代码
tensorflow rnn 最简单实现代码 #!/usr/bin/env python # -*- coding: utf-8 -*- import tensorflow as tf from te ...
- TensorFlow RNN MNIST字符识别演示快速了解TF RNN核心框架
TensorFlow RNN MNIST字符识别演示快速了解TF RNN核心框架 http://blog.sina.com.cn/s/blog_4b0020f30102wv4l.html
- Ubuntu16.04下搜狗输入法、Sublime Text 3的安装
Ubuntu16.04下搜狗输入法.Sublime Text 3的安装 一.搜狗输入法 1. 安装中文语言 默认在Ubuntu16.04下是没有中文的,需要安装中文,在System Settings- ...
- 论文列表——text classification
https://blog.csdn.net/BitCs_zt/article/details/82938086 列出自己阅读的text classification论文的列表,以后有时间再整理相应的笔 ...
随机推荐
- 【原创】python模拟腾讯网页登录
近日,研究Tencent网页模拟登录的过程,过程有些忐忑,但最终还是实现了这一功能.先将结果写于此,供大家参考: 其加密过程在c_login_old.js文件中执行,将JS关键代码提取出来如下: fu ...
- C++构造函数初始化列表与构造函数中的赋值的区别
C++类中成员变量的初始化有两种方式:构造函数初始化列表和构造函数体内赋值. 一.内部数据类型(char,int……指针等) class Animal { public: Animal(int wei ...
- 你不知道的react
1.props其实是JSX属性作为单个对象传递给组件:<Welcome name="john"> {name: 'john'} 2.ReactDOM.render()的 ...
- VMware内CentOS7虚拟机硬盘扩容
转自:https://blog.csdn.net/Wang_Xin_SH/article/details/77872885 简介 CentOS7虚拟机原硬盘空间只分配了10GB,需要扩容到20GB. ...
- 奇怪吸引子---LorenaMod2
奇怪吸引子是混沌学的重要组成理论,用于演化过程的终极状态,具有如下特征:终极性.稳定性.吸引性.吸引子是一个数学概念,描写运动的收敛类型.它是指这样的一个集合,当时间趋于无穷大时,在任何一个有界集上出 ...
- boost python3依赖安装
之前写过python dlib依赖的安装,当时是使用了系统默认的python2.7安装的boost.Python,如果自己编译好python3,想安装支持python3的boost.Python也是可 ...
- SQLSERVER 2012 之 “代理 XP”组件已作为此服务器安全配置的一部分被关闭
新建维护计划的时候遇到下图的报错信息 标题: Microsoft SQL Server Management Studio ------------------------------ “代理 XP” ...
- java 除法运算只保留整数位的3种方式
1.情景展示 根据提供的毫秒数进行除法运算,如果将毫秒数转换成小时,小时数不为0,则只取整数位,依此类推... 2.情况分析 可以使用3个函数实现 Math.floor(num) 只保留整数位 ...
- Rplidar学习(五)—— rplidar使用cartographer_ros进行地图云生成
一.Cartographer简介 Cartographer是google开源的通用2D和3D定位与地图同步构建的SLAM工具,并提供ROS接口.官网地址:https://github.com/goog ...
- 11g新特性 -- Virtual Private Catalogs
在oracle 11g之前版本中,如果用户只是需要访问recovery catalog 中的某个数据库,但是我们却只能将整个recovery catalog授权给他,无法做到严格限制. 现在,virt ...