<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[R.KING]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/darknightt</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; darknightt]]></copyright><item><title><![CDATA[[LeetcCode]Top 100 Liked Questions(1-10)]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/78466003</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/78466003</guid><author>darknightt</author><pubDate>Tue, 07 Nov 2017 11:27:07 +0800</pubDate><description><![CDATA[[LeetcCode]Top 100 Liked Questions(1-10)]]></description><category></category></item><item><title><![CDATA[神经网络与深度学习（三）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/71075703</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/71075703</guid><author>darknightt</author><pubDate>Mon, 01 May 2017 21:58:39 +0800</pubDate><description><![CDATA[卷积神经网络]]></description><category></category></item><item><title><![CDATA[neural networks学习笔记（二）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70892817</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70892817</guid><author>darknightt</author><pubDate>Fri, 28 Apr 2017 11:27:10 +0800</pubDate><description><![CDATA[继续看理论，补一下之前跳过的BP神经网络How the backpropagation algorithm works看了四分之一都在介绍概念和幺蛾子Hadamard product 
四个方程]]></description><category></category></item><item><title><![CDATA[neural networks学习笔记（一）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70786724</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70786724</guid><author>darknightt</author><pubDate>Wed, 26 Apr 2017 10:18:31 +0800</pubDate><description><![CDATA[improving the way neural networks learnthe cross-entropy cost function1.比较初始值不同的两个学习曲线，验证结论：当我们错得越厉害，学习得越快2.学习速度越来越慢的原因： 
sigmoid函数的导数在收敛到1或0的过程中越来越小趋近于0如何解决学习越来越慢？方法之一，用叉熵成本函数代替平方成本。1.如何理解它作为一个cost fu]]></description><category></category></item><item><title><![CDATA[和弦杂记]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70195718</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70195718</guid><author>darknightt</author><pubDate>Sun, 16 Apr 2017 15:24:08 +0800</pubDate><description><![CDATA[写给自己看，表述很随意]]></description><category></category></item><item><title><![CDATA[windows下xgboost的python配置]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70186432</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70186432</guid><author>darknightt</author><pubDate>Sat, 15 Apr 2017 19:43:26 +0800</pubDate><description><![CDATA[windows下安装xgboost，附编译好后python包]]></description><category></category></item><item><title><![CDATA[lasso（L1正则化、lasso回归）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70182902</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70182902</guid><author>darknightt</author><pubDate>Sat, 15 Apr 2017 10:17:59 +0800</pubDate><description><![CDATA[L1、L2正则化对比]]></description><category></category></item><item><title><![CDATA[Tikhonov regularization（L2正则化、岭回归）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70179848</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70179848</guid><author>darknightt</author><pubDate>Sat, 15 Apr 2017 01:23:02 +0800</pubDate><description><![CDATA[正则化公式推导]]></description><category></category></item><item><title><![CDATA[最小二乘法]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70177028</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70177028</guid><author>darknightt</author><pubDate>Sat, 15 Apr 2017 00:03:02 +0800</pubDate><description><![CDATA[最小二乘法公式推导]]></description><category></category></item><item><title><![CDATA[kaggle中的可视化（一）：House Prices]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70171916</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70171916</guid><author>darknightt</author><pubDate>Fri, 14 Apr 2017 14:31:09 +0800</pubDate><description><![CDATA[kaggle中预测的get started项目，原文链接。 
看原文可以入门特征工程，这里主要说可视化部分，用到matplotlib和seaborn。]]></description><category></category></item><item><title><![CDATA[pandas的绘图函数（一）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70171653</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70171653</guid><author>darknightt</author><pubDate>Fri, 14 Apr 2017 14:01:52 +0800</pubDate><description><![CDATA[依旧参照《利用Python进行数据分析》，过一遍基础。]]></description><category></category></item><item><title><![CDATA[树分类、线性回归和树回归的感性认知]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70169699</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70169699</guid><author>darknightt</author><pubDate>Fri, 14 Apr 2017 10:31:02 +0800</pubDate><description><![CDATA[假设一个数据集：
样本：n个人
特征：m个特征（体重、身高等）
类别：男，女
预测值：健康系数（范围0~100）]]></description><category></category></item><item><title><![CDATA[matplotlib学习记录（一）]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70161268</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70161268</guid><author>darknightt</author><pubDate>Thu, 13 Apr 2017 19:28:24 +0800</pubDate><description><![CDATA[先参照《利用Python进行数据分析》的内容过一遍matplotlib库基本api的操作，后续的学习添加一些fashion的东西，备忘用。]]></description><category></category></item><item><title><![CDATA[树回归]]></title><link>https://blog.youkuaiyun.com/darknightt/article/details/70154360</link><guid>https://blog.youkuaiyun.com/darknightt/article/details/70154360</guid><author>darknightt</author><pubDate>Thu, 13 Apr 2017 10:06:56 +0800</pubDate><description><![CDATA[为什么会引入树回归？因为现实生活中很多问题是非线性的，单纯的用线性回归得不到满意的结果。树回归的核心思想是数据集切分：既然全体数据集太复杂不好建模，就把它切分成很多份，每份方便建模就行了呗。至于怎么切分，切出来之后怎么建模就是树回归研究的问题。]]></description><category></category></item></channel></rss>