Link of the Paper: https://arxiv.org/pdf/1409.3215.pdf

Main Points:

  1. Encoder-Decoder Model: Input sequence -> A vector of a fixed dimensionality -> Target sequence.
  2. A multilayered  LSTM: The LSTM did not have difficulty on long sentences. Deep LSTMs significantly outperformed shallow LSTMs.
  3. Reverse Input: Better performance. While the authors do not have a complete explanation to this phenomenon, they believe that it is caused by the introduction of many short term dependencies to the dataset. LSTMs trained on reversed source sentences did much better on long sentences than LSTMs trained on the raw source sentences, which suggests that reversing the input sentences results in LSTMs with better memory utilization.

Other Key Points:

  1. A significant limitation: Despite their flexibility and power, DNNs can only be applied to problems whose inputs and targets can be sensibly encoded with vectors of fixed dimensionality.

Paper Reading - Sequence to Sequence Learning with Neural Networks ( NIPS 2014 )的更多相关文章

  1. Paper Reading - Deep Captioning with Multimodal Recurrent Neural Networks ( m-RNN ) ( ICLR 2015 ) ★

    Link of the Paper: https://arxiv.org/pdf/1412.6632.pdf Main Points: The authors propose a multimodal ...

  2. 【论文笔记】Learning Convolutional Neural Networks for Graphs

    Learning Convolutional Neural Networks for Graphs 2018-01-17  21:41:57 [Introduction] 这篇 paper 是发表在 ...

  3. PP: Sequence to sequence learning with neural networks

    From google institution; 1. Before this, DNN cannot be used to map sequences to sequences. In this p ...

  4. 《MATLAB Deep Learning：With Machine Learning，Neural Networks and Artificial Intelligence》选记

    一.Training of a Single-Layer Neural Network 1 Delta Rule Consider a single-layer neural network, as ...

  5. [C1W4] Neural Networks and Deep Learning - Deep Neural Networks

    第四周:深层神经网络(Deep Neural Networks) 深层神经网络(Deep L-layer neural network) 目前为止我们学习了只有一个单独隐藏层的神经网络的正向传播和反向 ...

  6. [C1W3] Neural Networks and Deep Learning - Shallow neural networks

    第三周:浅层神经网络(Shallow neural networks) 神经网络概述(Neural Network Overview) 本周你将学习如何实现一个神经网络.在我们深入学习具体技术之前,我 ...

  7. 目标检测--Scalable Object Detection using Deep Neural Networks(CVPR 2014)

    Scalable Object Detection using Deep Neural Networks 作者: Dumitru Erhan, Christian Szegedy, Alexander ...

  8. Sequence to Sequence Learning with Neural Networks论文阅读

    论文下载 作者(三位Google大佬)一开始提出DNN的缺点,DNN不能用于将序列映射到序列.此论文以机器翻译为例,核心模型是长短期记忆神经网络(LSTM),首先通过一个多层的LSTM将输入的语言序列 ...

  9. Paper Reading——LEMNA:Explaining Deep Learning based Security Applications

    Motivation: The lack of transparency of the deep  learning models creates key barriers to establishi ...

随机推荐

  1. Microsoft Visio / Project professional 2013 官方版本（下载）

    Microsoft Visio微软开发的一款软件, 它有助于 IT 和商务专业人员轻松地可视化.分析和交流复杂信息. 它能够将难以理解的复杂文本和表格转换为一目了然的 Visio 图表. 该软件通过创 ...

  2. Maven 高级应用

    Maven 的高级应用主要体现在 ==依赖==,==聚合==,==继承== * 依赖 就是在当前项目的pom.xml 总引入依赖的坐标 最最经常用到的 <dependencies> < ...

  3. Order by排序

    asc 升序(默认),desc 降序 order by 后面  可以加 列.表达式.别名.序号(从1开始) desc; --表达式 年薪 from emp order by 年薪 desc; --别名 ...

  4. iOS项目启动及启动时间优化

    app的启动入口Main函数: int main(int argc, char * argv[]) { @autoreleasepool { return UIApplicationMain(argc ...

  5. vue keep-alive 不生效 以及前进 后退 对数据刷新和保留缓存操作

    https://blog.csdn.net/sinat_37255207/article/details/89373825 因为项目Vue router 连续嵌套了好几层 首先检查keep-alive ...

  6. 添加一个js扩展方法

    String.prototype.repeatify=String.prototype.repeatify || function(times){ var str=''; for(var i=0;i& ...

  7. python if-elif-else 结构判断输入值处于何种年龄段

    输入变量 age 的值,再编写一个 if-elif-else 结构,根据 age的值判断处于人生的哪个阶段.如果一个人的年龄小于 2岁,就打印一条消息,指出他是婴儿.如果一个人的年龄为 2(含)-4岁 ...

  8. FROM_UNIXTIME/CONCAT

    将mysql查询结果中时间戳转化为时间格式 FROM_UNIXTIME( c.createtime, '%Y-%m-%d %H:%i:%S' ) 2个字段合并查询 CONCAT(d.`name`, ' ...

  9. 一图看懂JVM,JRE,JDK的关系

  10. 03.搭建Spark集群(CentOS7+Spark2.1.1+Hadoop2.8.0)

    接上一篇:https://www.cnblogs.com/yjm0330/p/10077076.html 一.下载安装scala 1.官网下载 2.spar01和02都建立/opt/scala目录,解 ...