<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[chiechie的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/chiechie</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; chiechie]]></copyright><item><title><![CDATA[Introduction to Boosted Trees]]></title><link>https://blog.youkuaiyun.com/chiechie/article/details/49334437</link><guid>https://blog.youkuaiyun.com/chiechie/article/details/49334437</guid><author>chiechie</author><pubDate>Thu, 22 Oct 2015 15:45:44 +0800</pubDate><description><![CDATA[none]]></description><category></category></item><item><title><![CDATA[随机优化工具 Stochastic Optimization Techniques]]></title><link>https://blog.youkuaiyun.com/chiechie/article/details/48802783</link><guid>https://blog.youkuaiyun.com/chiechie/article/details/48802783</guid><author>chiechie</author><pubDate>Tue, 29 Sep 2015 09:33:50 +0800</pubDate><description><![CDATA[@(Paper summaries)[Neural Networks|Optimization]

Stochastic Gradient Descent
Momentum
Nesterovs Accelerated Gradient
Adagrad
RMSProp
Adadelta
Adam
ESGD
Adasecant
vSGD
RpropNeu]]></description><category></category></item><item><title><![CDATA[git和Github常用操作]]></title><link>https://blog.youkuaiyun.com/chiechie/article/details/48802719</link><guid>https://blog.youkuaiyun.com/chiechie/article/details/48802719</guid><author>chiechie</author><pubDate>Tue, 29 Sep 2015 09:26:54 +0800</pubDate><description><![CDATA[git常用操作]]></description><category></category></item><item><title><![CDATA[Degrees of freedom]]></title><link>https://blog.youkuaiyun.com/chiechie/article/details/48174869</link><guid>https://blog.youkuaiyun.com/chiechie/article/details/48174869</guid><author>chiechie</author><pubDate>Wed, 02 Sep 2015 10:27:36 +0800</pubDate><description><![CDATA[Degrees of freedom in linear models[edit] 
The demonstration of the t and chi-squared distributions for one-sample problems above is the simplest example where degrees-of-freedom arise. However, simila]]></description><category></category></item><item><title><![CDATA[stepwise算法]]></title><link>https://blog.youkuaiyun.com/chiechie/article/details/47926257</link><guid>https://blog.youkuaiyun.com/chiechie/article/details/47926257</guid><author>chiechie</author><pubDate>Mon, 24 Aug 2015 00:32:42 +0800</pubDate><description><![CDATA[原始的交叉矩阵的q行q列进行sweep变换： 
 
可以看到sweep之后，原本叉乘矩阵的位置变成了它的逆矩阵，而第一行residueSS向量的第0个元素表示的是RSSqRSS_q,剩下的表示的是系数。 
如果我们想将q个变量扩充成q+1q+1个变量，可以参考下图： 
 
首先将ZZ分解为两部分：Z=Z1+Z2Z = Z_1+Z_2, 
Z1=(I−H)ZZ_1 = (I-H)Z，Z2=HZZ_2]]></description><category></category></item><item><title><![CDATA[A Tutorial on the SWEEP Operator–读书笔记]]></title><link>https://blog.youkuaiyun.com/chiechie/article/details/46803451</link><guid>https://blog.youkuaiyun.com/chiechie/article/details/46803451</guid><author>chiechie</author><pubDate>Wed, 08 Jul 2015 14:53:35 +0800</pubDate><description><![CDATA[算法原理对线性回归的最小二乘解，涉及到求矩阵的逆(XTX)−1(X^TX)^{-1}， 
 1.原始的矩阵求逆的方法–Gauss Jordan消去法，使用计算机来求解时，需要开辟另一个内存,存放变化时的右矩阵A−1A^{-1}. 
 2.sweep算法对高斯约旦消去法的一种改进，它可以节省存储空间，不需要开辟内存单独存放右矩阵。下面使用三阶例子来说明：（1）对于矩阵A,其扩展矩阵[A|I][A|I]]]></description><category></category></item></channel></rss>