<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[u200710的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/u200710</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; u200710]]></copyright><item><title><![CDATA[聚类]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107767939</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107767939</guid><author>u200710</author><pubDate>Fri, 07 Aug 2020 15:06:52 +0800</pubDate><description><![CDATA[Clustering]]></description><category></category></item><item><title><![CDATA[流形学习]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107765385</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107765385</guid><author>u200710</author><pubDate>Mon, 03 Aug 2020 16:39:21 +0800</pubDate><description><![CDATA[Manifold learning]]></description><category></category></item><item><title><![CDATA[高斯混合模型]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107762248</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107762248</guid><author>u200710</author><pubDate>Mon, 03 Aug 2020 15:38:31 +0800</pubDate><description><![CDATA[高斯混合模型
sklearn.mixture是一个能够学习高斯混合模型、抽样高斯模型和从数据中估计模型的包。同样，也提供了帮助决定正确组件数量的方法。
一个高斯混合模型是一个概率模型，它假设所有的数据点是从有限未知参数的高斯分布的混合生成的。可以将混合模型当作泛化的k均值聚类，以融合关于数据协方差和潜在高斯中心的信息。
高斯混合
GaussianMixture对象实现了expection-maximization算法来拟合高斯混合模型。它也能够得到多元模型的置信椭圆，计算贝叶斯信息准则来确定数据中聚集类别的]]></description><category></category></item><item><title><![CDATA[神经网络模型]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107729353</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107729353</guid><author>u200710</author><pubDate>Sat, 01 Aug 2020 13:53:50 +0800</pubDate><description><![CDATA[Neural Network Models (supervised)]]></description><category></category></item><item><title><![CDATA[概率校准]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107665193</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107665193</guid><author>u200710</author><pubDate>Wed, 29 Jul 2020 15:46:03 +0800</pubDate><description><![CDATA[Probability calibration]]></description><category></category></item><item><title><![CDATA[保序回归]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107664662</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107664662</guid><author>u200710</author><pubDate>Wed, 29 Jul 2020 15:10:45 +0800</pubDate><description><![CDATA[IsotonicRegression]]></description><category></category></item><item><title><![CDATA[半监督学习]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107662532</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107662532</guid><author>u200710</author><pubDate>Wed, 29 Jul 2020 14:44:44 +0800</pubDate><description><![CDATA[Semi-supervised learning]]></description><category></category></item><item><title><![CDATA[特征选择]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107605761</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107605761</guid><author>u200710</author><pubDate>Mon, 27 Jul 2020 16:54:27 +0800</pubDate><description><![CDATA[Feature Selection]]></description><category></category></item><item><title><![CDATA[多类别和多标签算法]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107593701</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107593701</guid><author>u200710</author><pubDate>Mon, 27 Jul 2020 10:13:26 +0800</pubDate><description><![CDATA[Multiclass and multilabel algorithms]]></description><category></category></item><item><title><![CDATA[集成方法]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107559244</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107559244</guid><author>u200710</author><pubDate>Sun, 26 Jul 2020 14:54:19 +0800</pubDate><description><![CDATA[Ensemble Methods: Bagging, Boosting, Stacked]]></description><category></category></item><item><title><![CDATA[决策树]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107535940</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107535940</guid><author>u200710</author><pubDate>Fri, 24 Jul 2020 13:31:48 +0800</pubDate><description><![CDATA[Decision Tree]]></description><category></category></item><item><title><![CDATA[朴素贝叶斯]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107535508</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107535508</guid><author>u200710</author><pubDate>Thu, 23 Jul 2020 14:22:26 +0800</pubDate><description><![CDATA[朴素贝叶斯
朴素贝叶斯是一系列监督学习算法，在每对特征值相互条件独立的假设上，使用贝叶斯定理。
不同的贝叶斯分类器主要在于其关于P(xi∣y)P(x_i | y)P(xi​∣y)的假设不同。尽管使用了这种过于简化的假设，朴素贝叶斯在文件分类和垃圾邮件过滤上应用广泛。他们仅需要很少的训练数据即可估计必要的参数。
Gaussian Naive Bayes
GaussianNB implements the Gaussian Naive Bayes algorithm for classification.
Mu]]></description><category></category></item><item><title><![CDATA[交叉分解]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107530854</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107530854</guid><author>u200710</author><pubDate>Thu, 23 Jul 2020 10:49:10 +0800</pubDate><description><![CDATA[Cross Decomposition]]></description><category></category></item><item><title><![CDATA[高斯过程]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107444485</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107444485</guid><author>u200710</author><pubDate>Thu, 23 Jul 2020 10:25:25 +0800</pubDate><description><![CDATA[高斯过程
高斯过程是为了求解回归和概率分类问题的一类通用的监督学习方法。其优点是

预测可以解释观测值
预测是概率的，这样可以计算经验信心区间
通用性

缺点是：

不是稀疏的，要使用全部样本和特征信息执行预测
在高维空间失效

高斯过程回归

...]]></description><category></category></item><item><title><![CDATA[最邻近]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107378111</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107378111</guid><author>u200710</author><pubDate>Sun, 19 Jul 2020 13:55:45 +0800</pubDate><description><![CDATA[最邻近(Nearest Neighbors)]]></description><category></category></item><item><title><![CDATA[随机梯度下降]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107331506</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107331506</guid><author>u200710</author><pubDate>Thu, 16 Jul 2020 10:40:54 +0800</pubDate><description><![CDATA[随机梯度下降
随机梯度下降是一种用于解决使用凸损失函数的线性分类器的方法，如，支持向量机和逻辑回归。]]></description><category></category></item><item><title><![CDATA[支持向量机]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107242709</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107242709</guid><author>u200710</author><pubDate>Tue, 14 Jul 2020 09:45:49 +0800</pubDate><description><![CDATA[支持向量机
支持向量机 (Support Vector Machines)是一系列用于分类、回归和异常点检测的监督学习方法。
支持向量机的优点是：

在高维数据空间的高效性
当维度远大于样本数量时，仍然有效
在决策函数中使用训练点的子集
通用性：对于决策函数，可以使用不同的核函数。提供了通用的核，但是也可以指定定制化的核

支持向量机的缺点是：

如果特征的数量远大于样本的数量时，为避免拟合，在选择核函数和正则项时非常关键
SVMs没有直接提供概率估计，需使用昂贵的五折交叉验证计算

在scikit-lea]]></description><category></category></item><item><title><![CDATA[核岭回归]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107220575</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107220575</guid><author>u200710</author><pubDate>Thu, 09 Jul 2020 12:20:56 +0800</pubDate><description><![CDATA[比较SVR和KRR的优劣]]></description><category></category></item><item><title><![CDATA[线性和二次判别分析]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107197509</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107197509</guid><author>u200710</author><pubDate>Wed, 08 Jul 2020 17:04:45 +0800</pubDate><description><![CDATA[线性和二次判别分析
线性判别分析(discriminant_analysis.LinearDiscriminantAnalysis)和二次判别分析(discriminant_analysis.QuadraticDiscriminantAnalysis)是两个经典带一次和二次决策面的分类器。
这些分类器因其有解析表达式、易计算、无超参的特点而格外吸引人。如下例：

...]]></description><category></category></item><item><title><![CDATA[线性模型]]></title><link>https://blog.youkuaiyun.com/u200710/article/details/107027763</link><guid>https://blog.youkuaiyun.com/u200710/article/details/107027763</guid><author>u200710</author><pubDate>Tue, 07 Jul 2020 12:20:23 +0800</pubDate><description><![CDATA[本文介绍线性模型中常见的模型和算法]]></description><category></category></item></channel></rss>