<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[geeksu的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/geeksu</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; geeksu]]></copyright><item><title><![CDATA[期望的塔性性质基础证明]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/156280822</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/156280822</guid><author>geeksu</author><pubDate>Fri, 26 Dec 2025 00:33:15 +0800</pubDate><description><![CDATA[EXEEX∣Y]]假设Y有界且离散Yy1​y2​y3​yi​1证明:令gYEX∣Yy: 给定y时, x的期望随着y的变化而变化EEX∣Y]]​EgY]]gy1​⋅PYy1​gy2​⋅PYy2​gy3​⋅PYy3​⋯gyi​⋅PYyi​y∑​EX∣Yyi​⋅PYyi​​又∵EX∣Yyi​。]]></description><category></category></item><item><title><![CDATA[dify部署问题：plugin launch error: init environment for plugin ... failed too many times]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/151051429</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/151051429</guid><author>geeksu</author><pubDate>Sun, 31 Aug 2025 21:22:59 +0800</pubDate><description><![CDATA[分析： 从报错日志上看，是网络引起的。而且是plugin_daemon-1容器引发的错误,下载“包”导致的错误，初步诊断应该是由于镜像仓库地址，默认使用了是pip官方镜像仓库地址，国内可能连不上或者连接不稳定。]]></description><category></category></item><item><title><![CDATA[一道题了解朴素贝叶斯]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/148503102</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/148503102</guid><author>geeksu</author><pubDate>Sun, 08 Jun 2025 00:56:15 +0800</pubDate><description><![CDATA[朴素贝叶斯（naive Bayes）法是一种基于概率的机器学习算法。它基于贝叶斯定理，并假设特征之间相互独立（这就是“朴素”的来源）。tip: 现实问题中，可能会遇到某些特征不存在的情况，这个时候需要使用到拉普拉斯平滑。学习与预测的效率都很高，是一种常用方法，在许多场景下表现得非常好，如文本分类（垃圾邮件检测）、情感分析等。下取得对应分类概率的最大值，最大值对应的分类即为所求。：表示分类的类别，在这里表示好果、坏果;: 先验概率，类Y的概率。的概率，对于每一个分类来说。: 后验概率，给定特征。]]></description><category></category></item><item><title><![CDATA[逻辑回归与Softmax]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/148484210</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/148484210</guid><author>geeksu</author><pubDate>Sat, 07 Jun 2025 00:11:31 +0800</pubDate><description><![CDATA[Softmax 函数是一种将任意实数向量转化为概率分布的函数，常用于多分类任务中作为神经网络或逻辑回归模型的输出层激活函数。它通过指数映射和归一化操作，确保输出向量的每个元素都在 (0, 1) 区间内，并且所有元素之和为 1。]]></description><category></category></item><item><title><![CDATA[关于逻辑回归的见解]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/148437901</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/148437901</guid><author>geeksu</author><pubDate>Thu, 05 Jun 2025 02:16:00 +0800</pubDate><description><![CDATA[逻辑回归通过线性回归输出映射到[0,1]区间表示类别概率，核心是Sigmoid函数（f(x)=1/(1+e^-x)），其导数具有f'(x)=f(x)(1-f(x))的特性。模型将线性结果输入Sigmoid函数得到概率值，通过阈值（如0.5）判断分类。损失函数采用对数损失（Log Loss），源自最大似然估计，对n个样本求平均负对数似然以优化模型。该设计保证损失标准化和优化稳定性，常用梯度下降求解最小值。]]></description><category></category></item><item><title><![CDATA[矩阵的偏导数]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/148401998</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/148401998</guid><author>geeksu</author><pubDate>Tue, 03 Jun 2025 15:41:08 +0800</pubDate><description><![CDATA[文章描述了，关于矩阵偏导数的基础推理，旨在补充AI学习过程中公式推理问题]]></description><category></category></item><item><title><![CDATA[hudi spark avro兼容性问题]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/111874692</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/111874692</guid><author>geeksu</author><pubDate>Mon, 28 Dec 2020 21:13:59 +0800</pubDate><description><![CDATA[在学习hudi过程中，参考了：https://hudi.apache.org/docs/quick-start-guide.html，在使用spark shell执行用例：
进入spark shell:


spark-shell \
--packages org.apache.spark:spark-avro_2.11:2.4.6,org.apache.avro:avro:1.8.2 \
--repositories http://maven.aliyun.com/nexus/content/groups]]></description><category></category></item><item><title><![CDATA[KNN]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78695523</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78695523</guid><author>geeksu</author><pubDate>Sat, 02 Dec 2017 14:31:58 +0800</pubDate><description><![CDATA[#_*_ coding:utf-8 _*_
from numpy import *
import operator
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.fontconfig_pattern as fontPattern
import os

#构造数据集
def createDataSet():]]></description><category></category></item><item><title><![CDATA[Decision Tree]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78678808</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78678808</guid><author>geeksu</author><pubDate>Thu, 30 Nov 2017 17:40:38 +0800</pubDate><description><![CDATA[1.使用sklearn来完成购买电脑的预测
import csv
from sklearn import preprocessing, tree

from sklearn.feature_extraction import DictVectorizer

featureList = []  # 特征向量容器  --> 注意这可是特征和特征值组成的dict所组成的List
labelList =]]></description><category></category></item><item><title><![CDATA[16.PCA续，奇异值分解SVD,独立成分分析ICA]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78491949</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78491949</guid><author>geeksu</author><pubDate>Thu, 09 Nov 2017 17:22:59 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[15.主成分析算法PCA]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78479456</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78479456</guid><author>geeksu</author><pubDate>Wed, 08 Nov 2017 15:38:52 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[14.混合高斯模型，混合贝叶斯模型，因子分析算法]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78463481</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78463481</guid><author>geeksu</author><pubDate>Tue, 07 Nov 2017 00:18:41 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[13.k-means聚类，混合高斯，EM算法]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78440906</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78440906</guid><author>geeksu</author><pubDate>Sat, 04 Nov 2017 00:28:50 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[12.贝叶斯正则化，在线学习,误差分析，销蚀分析]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78430310</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78430310</guid><author>geeksu</author><pubDate>Thu, 02 Nov 2017 22:43:27 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[11.VC维度，模型选择，特征选择]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78415606</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78415606</guid><author>geeksu</author><pubDate>Wed, 01 Nov 2017 16:33:27 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[10.误差理论]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78405783</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78405783</guid><author>geeksu</author><pubDate>Tue, 31 Oct 2017 17:23:42 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[9.非线性支持向量机与核函数]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78394831</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78394831</guid><author>geeksu</author><pubDate>Mon, 30 Oct 2017 17:01:46 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[8.线性可分支持向量机(SVM)]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78317988</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78317988</guid><author>geeksu</author><pubDate>Mon, 23 Oct 2017 13:43:47 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[7.朴素贝叶斯的多项式事件模型，SVM初步，神经网络初步]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78216355</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78216355</guid><author>geeksu</author><pubDate>Thu, 12 Oct 2017 16:25:13 +0800</pubDate><description><![CDATA[]]></description><category></category></item><item><title><![CDATA[spark更新广播变量实例]]></title><link>https://blog.youkuaiyun.com/geeksu/article/details/78199164</link><guid>https://blog.youkuaiyun.com/geeksu/article/details/78199164</guid><author>geeksu</author><pubDate>Wed, 11 Oct 2017 00:27:48 +0800</pubDate><description><![CDATA[实际项目中需要使用更新广播变量的方式来满足需求，其代码具体如下：
@NotProguard
public class WbBroadcastWrapper {
	
	Broadcast> broadcastBlackWhiteList = null;
	private static WbBroadcastWrapper obj = new WbBroadcastWrapper();

    p]]></description><category></category></item></channel></rss>