Retrofitting Analysis

To figure out the process of retrofitting[1] objective updating, we do the following math.

Forward Derivation

\[
\psi(Q) = \sum_{i=1}^{n}\left[ \alpha_i||q_i-\hat{q_i}||^2 + \sum\beta||q_i-q_j||^2 \right] \\
\frac{\partial \psi(Q)}{\partial q_i} = \alpha_i(q_i-\hat{q_i}) + \sum\beta(q_i-q_j) = 0 \\
(\alpha_i+\sum\beta_{ij})q_i -\alpha_i\hat{q_i} -\sum\beta_{ij}q_j = 0 \\
q_i = \frac{\sum\beta_{ij}q_j+\alpha_i\hat{q_i}}{\sum\beta_{ij}+\alpha_i}
\]

Backward Derivation

This is how I understood this updating equation.

In the paper[1], it has mentioned "We take the first derivative of \(\psi\) with respect to one qi vector, and by equating it to zero", hence we get follow idea:
\[
\frac{\partial\psi(Q)}{\partial q_i} = 0
\]

And,

\[
q_i = \frac{\sum\beta_{ij}q_j+\alpha_i\hat{q_i}}{\sum\beta_{ij}+\alpha_i} \\
\alpha_iq_i - \alpha_i\hat{q_j} + \sum\beta_{ij}q_i - \sum\beta q_j = 0 \\
\alpha_i(q_i-\hat{q_j})+ \sum\beta_{ij}(q_i-q_j) = 0
\]

Apparently,
\[
\frac{\partial\psi(Q)}{\partial q_i} = \alpha_i(q_i-\hat{q_j})+ \sum\beta_{ij}(q_i-q_j) = 0
\]

Reference

Faruqui M, Dodge J, Jauhar S K, et al. Retrofitting Word Vectors to Semantic Lexicons[J]. ACL, 2015.

Retrofitting Analysis的更多相关文章

  1. IJCAI 2019 Analysis

    IJCAI 2019 Analysis 检索不到论文的关键词:retrofitting word embedding Getting in Shape: Word Embedding SubSpace ...

  2. Why many EEG researchers choose only midline electrodes for data analysis EEG分析为何多用中轴线电极

    Source: Research gate Stafford Michahial EEG is a very low frequency.. and literature will give us t ...

  3. Automated Memory Analysis

    catalogue . 静态分析.动态分析.内存镜像分析对比 . Memory Analysis Approach . volatility: An advanced memory forensics ...

  4. Sentiment Analysis resources

    Wikipedia: Sentiment analysis (also known as opinion mining) refers to the use of natural language p ...

  5. Call for Papers IEEE/ACM International Conference on Advances in Social Network Analysis and Mining (ASONAM)

    IEEE/ACM International Conference on Advances in Social Network Analysis and Mining (ASONAM) 2014 In ...

  6. 主成分分析(principal components analysis, PCA)

    原理 计算方法 主要性质 有关统计量 主成分个数的选取 ------------------------------------------------------------------------ ...

  7. 《利用Python进行数据分析: Python for Data Analysis 》学习随笔

    NoteBook of <Data Analysis with Python> 3.IPython基础 Tab自动补齐 变量名 变量方法 路径 解释 ?解释, ??显示函数源码 ?搜索命名 ...

  8. Python for Data Analysis

    Data Analysis with Python ch02 一些有趣的数据分析结果 Male描述的是美国新生儿男孩纸的名字的最后一个字母的分布 Female描述的是美国新生儿女孩纸的名字的最后一个字 ...

  9. 使用SQL Server Analysis Services数据挖掘的关联规则实现商品推荐功能(七)

    假如你有一个购物类的网站,那么你如何给你的客户来推荐产品呢?这个功能在很多电商类网站都有,那么,通过SQL Server Analysis Services的数据挖掘功能,你也可以轻松的来构建类似的功 ...

随机推荐

  1. Spring 资源加载

    pom.xml ``` org.springframework spring-core 4.3.14.RELEASE org.springframework spring-beans 4.3.16.R ...

  2. java8学习之自定义收集器实现

    在上次花了几个篇幅对Collector收集器的javadoc进行了详细的解读,其涉及到的文章有: http://www.cnblogs.com/webor2006/p/8311074.html htt ...

  3. 【bzoj 4025 改编版】graph

    题意 给定一张 \(n\) 个点 \(m\) 条边的无向图,问删去每个点后,原图是不是二分图.输出一个长度为 \(n\) 的 \(\text{01}\) 串表示答案. 多组数据. \(T\le 5,\ ...

  4. @Mapper和@Repository的区别

    参考博客地址 https://www.cnblogs.com/wangshen31/p/8735037.html 相同点 两个都是注解在Dao上 不同 @Repository需要在Spring中配置扫 ...

  5. HTTP与TCP的区别和联系(转)

    https://www.cnblogs.com/baizhanshi/p/8482612.html

  6. python 异常处理(五)

    异常处理&异常基类 1.处理异常 try.....except 语法: 1) try: 放可能会出现问题的代码 except: 处理错误的方式 例如: try: print(ab)  #无错执 ...

  7. 在JavaScript中,++在前和++在后有什么区别

    一.++可以与输出语句写在一起,++写在变量前和写在变量后不是一个意思++ i 和 i ++ 区别在于运算顺序和结合方向. 在JavaScript中有两种自加运算,其运算符均为 ++,功能为将运算符自 ...

  8. POJ - 2774 Long Long Message (后缀数组/后缀自动机模板题)

    后缀数组: #include<cstdio> #include<algorithm> #include<cstring> #include<vector> ...

  9. 我说CMMI之六:CMMI的评估--转载

    版权声明:本文为博主原创文章,遵循 CC 4.0 BY-SA 版权协议,转载请附上原文出处链接和本声明.本文链接:https://blog.csdn.net/dylanren/article/deta ...

  10. restful api 相关

    404:资源没有找到400:参数错误 200:Get获取成功201:Post创建成功202:Put更新成功 401:未授权403:当前的资源禁止 500:服务器的未知错误 错误码 错误信息 当前url ...