
自然语言处理
文章平均质量分 93
qq_长期不在
这个作者很懒,什么都没留下…
展开
-
Distributed Representations of Sentences and Documents(Doc2Vec简摘)
Distributed Representations of Sentences and Documents(Doc2Vec简摘)介绍AlgorithmsParagraph Vector: A distributed memory model优点Many machine learning algorithms require the input to be represented as a fixed-length feature vector. When it comes to texts, one o原创 2021-02-04 18:48:07 · 439 阅读 · 0 评论 -
Deep Unordered Composition Rivals Syntactic Methods for Text Classification(简摘DAN模型)
Deep Unordered Composition Rivals Syntactic Methods for Text Classification(简摘)摘要成果结论摘要Many existing deep learning models for natural language processing tasks focus on learning the compositionality of their inputs, which requires many expensive computat原创 2021-02-02 11:25:40 · 742 阅读 · 0 评论 -
深度学习文本分类文献综述(翻译自Deep Learning Based Text Classification: A Comprehensive Review)
深度学习文本分类文献综述摘要介绍1. 文本分类任务2.文本分类中的深度模型2.1 Feed-Forward Neural Networks2.2 RNN-Based Models2.3 CNN-Based Models2.4 Capsule Neural Networks2.5 Models with Attention Mechanism2.6 Memory-Augmented Networks2.7 Transformers2.8 Graph Neural Networks2.9 Siamese Neu原创 2021-02-01 19:16:02 · 5707 阅读 · 0 评论 -
tf.nn.sampled_softmax_loss用法简单介绍
tf.nn.sampled_softmax_loss用法简单介绍在研究Skip-gram模型时遇到了采用方式的softmax,一时没有搞明白,下面做个小案例试一下。tf.nn.sampled_softmax_loss函数的参数类别如下:tf.nn.sampled_softmax_loss( weights, # Shape (num_classes, dim) 神经网络输出层权重 biases,原创 2020-10-19 19:03:31 · 3382 阅读 · 0 评论 -
Distributed Representations of Words and Phrases and their Compositionality 论文简解
Distributed Representations of Words and Phrases and their Compositionality 论文简解摘要1.介绍2. Skip-gram 模型2.1 Hierarchical Softmax2.2 Negative Sampling2.3 Subsampling of Frequent Words3. 测试结果4. 总结摘要该文章的主要研究工作是在Skip-gram模型的基础上提出了几种模型的改进方法,可以提升原始Skip-gram模型的训练速原创 2020-10-12 15:32:55 · 646 阅读 · 0 评论 -
词性标注HMM模型之Feature-Rich Part-of-Speech Tagging with a Cyclic Dependency Network (2003)论文解读
这里写目录标题概述概述原创 2020-05-23 15:39:49 · 669 阅读 · 0 评论 -
词性标注HMM模型之TnT — A Statistical Part-of-Speech Tagger (2000) 论文解读
这里写目录标题概述内容背景介绍模型体系架构目标函数推导Smoothing 处理未知词的处理大小写问题定向搜索总结概述该文作者是德国萨尔大学的Thorsten Brants,作者认为Tri-Grams’n’Tages(TnT)可以作为一种有效的词性标记统计方法,并且作者认为基于马尔科夫(Markov model)的标记器理论上性能是和其他模型相当的,此外使用平滑的方法(Smoothing)处理了未出现词的概率。下面我们就来看看该文作者具体是怎么实现的。内容背景介绍语言处理系统都会使用词性标记器进行预处原创 2020-05-21 14:30:53 · 814 阅读 · 0 评论