![](https://img-blog.csdnimg.cn/20201014180756922.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
论文笔记
文章平均质量分 69
长安居不易
Less dating, more coding
展开
-
KV-MemNN
论文出处:Key-Value Memory Networks for Directly Reading Documents代码(lua+torch):https://github.com/facebook/MemNN/tree/master/KVmemnnWikiMovies数据集:https://research.fb.com/downloads/babi/1.Introduction原创 2017-07-26 21:24:34 · 1892 阅读 · 0 评论 -
Dynamic Memory Networks & DMN+
DMN出自论文Ask Me Anything: Dynamic Memory Networks for Natural Language Processing(2015-06),这个模型用于Question Answering并且可以拓展到情感分析和词性标注。正如Introduce中所提到的,很多NLP问题都可以看作QA问题,比如机器翻译、命名实体标注、文本分类问题如情感分析、共指消解(corefe原创 2017-07-25 14:36:38 · 5074 阅读 · 1 评论 -
Dynamic Memory Tensor Networks笔记
DMTN是对DMN的改进,借鉴了NTN(Neural Tensor Networks)中的相似度评分函数,在bAbI-1k数据集上比DMN的结果要好。模型出自论文Ask Me Even More: Dynamic Memory Tensor Networks(Extended Model), theano实现代码:https://github.com/rgsachin/DMTNDMTN主要改原创 2017-07-24 10:17:01 · 819 阅读 · 0 评论 -
Recurrent Entity Network笔记
Recurrent Entity Network笔记Recurrent Entity Network简称EntNet,最初在论文TRACKING THE WORLD STATE WITH RECURRENT ENTITY NETWORKS中提出,文中给出了lua+torch的代码地址。作者包括Facebook AI研究院的Mikael Henaff, Jason Weston(MemNN和MemN原创 2017-07-23 17:19:40 · 2446 阅读 · 1 评论 -
Relation Network笔记
论文:A simple neural network module for relational reasoninggithub代码: https://github.com/siddk/relation-network1.Introduction智能行为很重要的一点就是能够推理实体间的关系。这一点CNN和MLP都做的不好。这篇文章提出了Relation Networks(RNs), 并且在vis原创 2017-08-16 16:59:55 · 10388 阅读 · 3 评论 -
Aspect Level Sentiment Classification with Deep Memory Network笔记
论文: Aspect Level Sentiment Classification with Deep Memory Network 代码:https://github.com/ganeshjawahar/mem_absa这篇文章主要针对aspect level sentiment analysis. 借鉴了MemN2N,在Sem-Eval的laptop和restaurant两种数据上,跑出的结果原创 2017-10-07 16:09:24 · 2162 阅读 · 1 评论 -
Detection Algorithms for Communication Systems Using Deep Learning笔记
论文:Detection Algorithms for Communication Systems Using Deep Learning 作者:Nariman Farsad, Andrea Goldsmith这篇文章是在知乎上看到链接的,所以就找过来瞅了一下,还挺有意思。文中说传统的信道建模,在已知先验概率PX(x)P_X(x)的情况下,通过计算最大后验概率(MAP)来解码,即xk=argma原创 2017-10-16 14:49:19 · 925 阅读 · 1 评论