Low-rank approximations

Give matrix and a positive integer , we wish to find an matrix of rank at most ，so as to minimize the Frobenius norm of the matrix difference ，defined to be

Thus, the Frobenius norm of measures the discrepancy between and ; our goal is to find a matrix that minimizes this discrepancy, while constraining to have rank at most . If is the rank of , clearly and the Frobenius norm of the discrepancy is zero in this case. When is far smaller than , we refer to as a low-rank approximation.

SVD

  1. Given , constuct its SVD int the form of .
  2. Derive from the matrix formed by replacing by zeros the smallest sigular values on the diagnal of .
  3. Compute and output as the rank- approximation to .

http://nlp.stanford.edu/IR-book/html/htmledition/low-rank-approximations-1.html

Low-rank approximations的更多相关文章

  1. Generalized Low Rank Approximation of Matrices

    Generalized Low Rank Approximations of Matrices JIEPING YE*jieping@cs.umn.edu Department of Computer ...

  2. 吴恩达机器学习笔记59-向量化：低秩矩阵分解与均值归一化（Vectorization: Low Rank Matrix Factorization & Mean Normalization）

    一.向量化:低秩矩阵分解 之前我们介绍了协同过滤算法,本节介绍该算法的向量化实现,以及说说有关该算法可以做的其他事情. 举例:1.当给出一件产品时,你能否找到与之相关的其它产品.2.一位用户最近看上一 ...

  3. Sparse Principal Component Analysis via Regularized Low Rank Matrix Approximation(Adjusted Variance)

    目录 前言 文章概述 固定\(\widetilde{\mathrm{v}}\) 固定\(\widetilde{\mathrm{u}}\) Adjusted Variance 前言 这篇文章用的也是交替 ...

  4. 推荐系统(recommender systems):预测电影评分--构造推荐系统的一种方法：低秩矩阵分解(low rank matrix factorization)

    如上图中的predicted ratings矩阵可以分解成X与ΘT的乘积,这个叫做低秩矩阵分解. 我们先学习出product的特征参数向量,在实际应用中这些学习出来的参数向量可能比较难以理解,也很难可 ...

  5. 矩阵分解（rank decomposition）文章代码汇总

    矩阵分解(rank decomposition)文章代码汇总 矩阵分解(rank decomposition) 本文收集了现有矩阵分解的几乎所有算法和应用,原文链接:https://sites.goo ...

  6. <<Numerical Analysis>>笔记

    2ed,  by Timothy Sauer DEFINITION 1.3A solution is correct within p decimal places if the error is l ...

  7. [综述]Deep Compression/Acceleration深度压缩/加速/量化

    Survey Recent Advances in Efficient Computation of Deep Convolutional Neural Networks, [arxiv '18] A ...

  8. <Numerical Analysis>(by Timothy Sauer) Notes

    2ed,  by Timothy Sauer DEFINITION 1.3A solution is correct within p decimal places if the error is l ...

  9. 网络压缩论文集(network compression)

    Convolutional Neural Networks ImageNet Models Architecture Design Activation Functions Visualization ...

  10. CVPR 2015 papers

    CVPR2015 Papers震撼来袭! CVPR 2015的文章可以下载了,如果链接无法下载,可以在Google上通过搜索paper名字下载(友情提示:可以使用filetype:pdf命令). Go ...

随机推荐

  1. 对于python这门课程的一些想法、计划、期望

    本人是一名大二的码农,专业信息安全.之前在知乎上看到过对于python一些评论,说用python写的代码和诗一样.也在网上大概的了解了一下,python要求有严格的缩进.学习python语言,最想学的 ...

  2. Java高级篇（三）——JDBC数据库编程

    JDBC是连接数据库和Java程序的桥梁,通过JDBC API可以方便地实现对各种主流数据库的操作.本篇将介绍一下如何使用JDBC操作数据库(以MySQL为例). 一.JDBC JDBC制定了统一访问 ...

  3. MySQL/MariaDB触发器

    本文目录:1.创建触发器2.insert触发器3.delete触发器4.update触发器5.通过on duplicate key update分析触发器触发原理6.replace to算法验证7.查 ...

  4. opencv2.4.9卸载安装

    1.安装opencv2.4.9过程中遇到的问题. 1.使用 sudo apt-get install libopencv-dev 安装OpenCV 2.4.9 发现版本为2.4.8,故卸载重装. 查看 ...

  5. [LeetCode] Pour Water 倒水

    We are given an elevation map, heights[i] representing the height of the terrain at that index. The ...

  6. [LeetCode] Construct String from Binary Tree 根据二叉树创建字符串

    You need to construct a string consists of parenthesis and integers from a binary tree with the preo ...

  7. javaApplication中如何使用log4j

  8. ML笔记：Deep Learning

    非DL:要找好的特征 DL:无需找好的特征,但新问题:要设计好的网络架构

  9. bzoj 3745: [Coci2015]Norma

    Description Solution 考虑分治: 我们要统计跨越 \(mid\) 的区间的贡献 分最大值和最小值所在位置进行讨论: 设左边枚举到了 \(i\),左边 \([i,mid]\) 的最大 ...

  10. 【Luogu P2709 小B的询问】莫队

    题目描述 小B有一个序列,包含N个1~K之间的整数.他一共有M个询问,每个询问给定一个区间[L..R],求Sigma(c(i)^2)的值,其中i的值从1到K,其中c(i)表示数字i在[L..R]中的重 ...