目录

* Recap <https://www.cnblogs.com/nickchen121/p/10961001.html#recap>
* Sentiment Analysis
<https://www.cnblogs.com/nickchen121/p/10961001.html#sentiment-analysis>
* Proposal <https://www.cnblogs.com/nickchen121/p/10961001.html#proposal>
* S1.Weight sharing
<https://www.cnblogs.com/nickchen121/p/10961001.html#s1.weight-sharing>
* Naive version
<https://www.cnblogs.com/nickchen121/p/10961001.html#naive-version>
* Weight share
<https://www.cnblogs.com/nickchen121/p/10961001.html#weight-share>
* S2.Consistent memory
<https://www.cnblogs.com/nickchen121/p/10961001.html#s2.consistent-memory>
* Unfolded model
<https://www.cnblogs.com/nickchen121/p/10961001.html#unfolded-model>
* Formulation
<https://www.cnblogs.com/nickchen121/p/10961001.html#formulation>
* Overall Diagram
<https://www.cnblogs.com/nickchen121/p/10961001.html#overall-diagram>
* One more thing
<https://www.cnblogs.com/nickchen121/p/10961001.html#one-more-thing>
* How to Train?
<https://www.cnblogs.com/nickchen121/p/10961001.html#how-to-train>
Recap



Sentiment Analysis



Proposal



* Long sentence
* 100 + words
* too much parameters[\(w_N,b_N\)]
* No context information
* consistent tensor
S1.Weight sharing

* 类似于卷积的卷积核,卷积视野,权重分享


Naive version



Weight share

* 权重分享


S2.Consistent memory

* 需要一个东西存储语境,也就是每个单词的语境信息


* 不断地对语境信息进行堆叠得到h5,直接使用h5作为判断标准


Unfolded model



* 在时间轴上折叠,不断地更新h


Formulation

* 通过激活函数tanh不断地叠加上个时间戳的信息


Overall Diagram

* 全览分解图


One more thing



How to Train?


友情链接
KaDraw流程图
API参考文档
OK工具箱
云服务器优惠
阿里云优惠券
腾讯云优惠券
华为云优惠券
站点信息
问题反馈
邮箱:[email protected]
QQ群:637538335
关注微信