2017-06-01 Lu Huang
之前简单介绍过 LSTM,网址,本文将利用 LSTM 来实现一个文本情感分类模型,是基于 Keras 和 Python 的。 概要 本文以词为单位,先分词,然后将每个句子截断为100词(不够则补空字符串),然后将句子以“词-词向量(embedding)”的矩阵形式输入到 LSTM 模型中进行学习分类。 本文使用的语料和参考来源于 文本情感分类(三):分词 OR 不分词. 数据预处理 数据读取与分词 ```python 读取数据并分词 pos = pd.read_excel('pos.xls', header=None) pos['label'] = 1 neg = pd.read_excel('neg.xls', he 继续阅读 »
2017-05-31 Lu Huang
之前简单介绍过 LSTM,网址,本文将利用 LSTM 构建一个故事生成的模型。 数据及其预处理 数据源 我们使用的数据是 Aesop’s Fables 里的一部分内容,如下: long ago , the mice had a general council to consider what measures they could take to outwit their common enemy , the cat . some said this , and some said that but at last a young mouse got up and said he had a proposal to make 继续阅读 »
2017-05-30 Lu Huang
This blog is reprinted from colah's blog and some changes are added by myself. About RNN Humans don’t start their thinking from scratch every second. And traditional neural networks have a major shortcoming, and they cannot learn from the previous information. Recurrent neural networks (RNN) address this issue. RNN 继续阅读 »
2016-11-09 安兴乐-siler
Text Recognition Fully Convolutional Recurrent Network for Handwritten Chinese Text Recognition paper Deep LSTM Networks for Online Chinese Handwriting Recognition 2016 ICFHR Convolution Multi-directional Recurrent Network for Offline Handwritten Text Recognition 2016 ICFHR Sequence to Sequen 继续阅读 »