<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[MLee的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/Recall_Tomorrow</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; Recall_Tomorrow]]></copyright><item><title><![CDATA[golang并发（同步）多任务高性能执行&聚合]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/138906609</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/138906609</guid><author>Recall_Tomorrow</author><pubDate>Wed, 15 May 2024 14:16:19 +0800</pubDate><description><![CDATA[golang并发高性能执行多任务，并聚合多任务执行结果与执行状态]]></description><category></category></item><item><title><![CDATA[RNN(LSTM)数据形式及Padding操作处理变长时序序列dynamic_rnn]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/89852099</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/89852099</guid><author>Recall_Tomorrow</author><pubDate>Sun, 05 May 2019 17:54:01 +0800</pubDate><description><![CDATA[Summary

RNN

样本一样，计算的状态值和输出结构一致，也即是说只要当前时刻的输入值也前一状态值一样，那么其当前状态值和当前输出结果一致，因为在当前这一轮训练中权重参数和偏置均未更新
RNN的最终状态值与最后一个时刻的输出值一致
输入数据要求格式为，shape=(batch_size, step_time_size, input_size),那么，state的shape=(batch_s...]]></description><category></category></item><item><title><![CDATA[TensorFlow中髙维张量乘法运算（RNN批量序列损失计算）]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/89851079</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/89851079</guid><author>Recall_Tomorrow</author><pubDate>Sun, 05 May 2019 17:27:55 +0800</pubDate><description><![CDATA[髙维张量tensor之间的乘法，均类似于二维矩阵的乘法，但需要满足最后两个维度符合矩阵乘法要求的维度，且其他维度均必须要相等的原则
import tensorflow as tf
import numpy as np

a = tf.constant(np.arange(1, 13, dtype=np.float32), shape=[2, 2, 3])
b = tf.constant(np.ar...]]></description><category></category></item><item><title><![CDATA[python实现单链表中常见功能]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/88873933</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/88873933</guid><author>Recall_Tomorrow</author><pubDate>Thu, 28 Mar 2019 19:46:50 +0800</pubDate><description><![CDATA[单链表实现code




序号
方法
描述




0
ll = LinkList(seq)
创建一个单链表对象


1
LinkList.size_()
链表中当前元素的个数


2
LinkList.is_empty()
判断链表是否为空


3
LinkList.clear()
将链表清空


4
LinkList.append(obj)
在链表末尾添加一个新的对象


5
LinkLis...]]></description><category></category></item><item><title><![CDATA[基于矩阵分解的推荐算法]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/80218051</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/80218051</guid><author>Recall_Tomorrow</author><pubDate>Sun, 06 May 2018 22:07:33 +0800</pubDate><description><![CDATA[推荐系统

&amp;amp;nbsp;&amp;amp;nbsp;&amp;amp;nbsp;&amp;amp;nbsp;&amp;amp;nbsp;&amp;amp;nbsp;&amp;amp;nbsp;&amp;amp;nbsp;\ \ \ \ 我们知道，协同过滤算法（Collaborative Filtering）是最常用的推荐算法，其就是通过从用户（User）的历史行为数据中挖掘出用户的兴趣，并为其推荐可能感兴趣的项（Item），常用的方法有，基于用户的协同过滤算法和基于项的协同过滤算法。 
&a]]></description><category></category></item><item><title><![CDATA[决策树算法原理及实现]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79553514</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79553514</guid><author>Recall_Tomorrow</author><pubDate>Wed, 14 Mar 2018 16:33:32 +0800</pubDate><description><![CDATA[欢迎大家查看实现的完整代码。。。



决策树模型

&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;\ \ \ \ 分类决策树模型是一种描述对实例进行分类的树形结构。决策树由结点和有向边组成。结点有两种类型：内部节点和叶节点，内部节点表示一个特征或属性，叶节点表示一个分类类别。 
&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbs]]></description><category></category></item><item><title><![CDATA[优化问题之拉格朗日乘子法&KKT条件分析]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79515824</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79515824</guid><author>Recall_Tomorrow</author><pubDate>Sun, 11 Mar 2018 13:51:59 +0800</pubDate><description><![CDATA[优化问题， 拉格朗日函数


无约束优化问题

minf(x)minf(x)\min f(x)，由Fermat’s theorem可知，可微函数的极值点都是其驻点（必要条件），故令其导数为零即可求解，当然也可利用梯度下降算法求解；
等式约束优化问题

minf0(x),&amp;nbsp;&amp;nbsp;s.t.,&amp;nbsp;hi(x)=0,&amp;nbsp;i=1,2,⋯,pminf0(x),&amp;nbsp;&amp;nbsp;s.t.,&amp;...]]></description><category></category></item><item><title><![CDATA[文本表示(Text Representation)之词集模型(SOW)&词袋模型(BOW)&TF-IDF模型]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79488639</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79488639</guid><author>Recall_Tomorrow</author><pubDate>Thu, 08 Mar 2018 20:41:04 +0800</pubDate><description><![CDATA[转载请注明来源  http://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79488639  
欢迎大家查看这些模型简单实现的代码…… 
&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;\ \ \ \ 对于一个包含若干个文档的语料库(Corpus)C={doc1,doc2,⋯,docm}C={doc1...]]></description><category></category></item><item><title><![CDATA[聚类算法之K-Means算法]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79456662</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79456662</guid><author>Recall_Tomorrow</author><pubDate>Tue, 06 Mar 2018 12:50:35 +0800</pubDate><description><![CDATA[聚类    \ \ \ \ 对大量未进行标注（Unsupervised learning）的数据集，并按照数据集本身的内部数据特征将数据划分为多个不同的类别，从而使得类别内的数据相似度高，而类别间的数据差异性较大。因此，它的重点就在于计算样本间的相似度（Similarity）。k-means算法思想与原理    \ \ \ \ 输入算法所需数据，数据样本集D={X1,X2,⋯,Xm}\mathcal]]></description><category></category></item><item><title><![CDATA[泰坦尼克之灾(逻辑回归应用)]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79420581</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79420581</guid><author>Recall_Tomorrow</author><pubDate>Fri, 02 Mar 2018 18:40:40 +0800</pubDate><description><![CDATA[问题背景及其描述

&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;\ \ \ \ 背景描述戳这里主要讲述了，在泰坦尼克号的首航中，不幸与冰山相撞，导致大量乘客与船员伤亡，而没有足够的救生艇是造成这次悲剧的主要原因之一。虽说那些幸存者是比较幸运的，但很明显对于妇女、孩子以及上层阶级等这些群体的人是更可能幸存下来的。So，我们的任务就是根据提供的初始数...]]></description><category></category></item><item><title><![CDATA[简述LDA,PCA,SVD原理推导及其简单应用]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79408355</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79408355</guid><author>Recall_Tomorrow</author><pubDate>Thu, 01 Mar 2018 17:12:33 +0800</pubDate><description><![CDATA[线性判别分析(Linear Discriminant Analysis)    \ \ \ \  对于给定的标记数据(xi,ci),其中xi∈Rn,ci∈{0,1}(x_i, c_i), 其中x_i\in \mathcal{R^n},c_i\in \{0, 1\}，此时设计一个分类器(Classifier)，将这两类数据分开，另外这两类数据线性可分(存在一个超平面Σ1\mathcal{\Sigma_]]></description><category></category></item><item><title><![CDATA[信息论基本概念(熵、联合熵、条件熵、相对熵、互信息)讲述与推导]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79088414</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/79088414</guid><author>Recall_Tomorrow</author><pubDate>Wed, 17 Jan 2018 18:06:44 +0800</pubDate><description><![CDATA[熵（Entropy）

​   熵是随机变量的不确定性的度量。设&#xA0;X&#xA0;" role="presentation"> X  X \ X\ 是一个离散的随机变量，其取值空间为&#xA0;X&#xA0;" role="presentation"> X  X \ \mathcal{X}\ ，概率密度函数p(x)=P(X=x),x&#x2208;X" role="presentation]]></description><category></category></item><item><title><![CDATA[使用朴素贝叶斯进行文档分类]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78912688</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78912688</guid><author>Recall_Tomorrow</author><pubDate>Wed, 27 Dec 2017 15:22:23 +0800</pubDate><description><![CDATA[朴素贝叶斯文档分类github代码
朴素(naive)贝叶斯之朴素 
 这里的朴素一词是指，即一个特征或者某个词条(token)的出现的可能性与它和其他单词的相邻性没有关系(当然，这个假设是不合理的，就如水果一词出现在健康一词周围的概率就明显高于出现在不健康一词周围的概率)，另外需要补充的是，这里的每一个特征也假设是同等重要的(也就是说每一个词条只考虑出现与否(set-of-words model)]]></description><category></category></item><item><title><![CDATA[matplotlib简单教程]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78896951</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78896951</guid><author>Recall_Tomorrow</author><pubDate>Mon, 25 Dec 2017 21:11:38 +0800</pubDate><description><![CDATA[Simple plot

In this section, we want to draw the cosine and sine functions on the same plot. Starting from the default settings, we’ll enrich the figure step by step to make it nicer.



import numpy...]]></description><category></category></item><item><title><![CDATA[机器学习常用数学公式（梯度说明）]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78896872</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78896872</guid><author>Recall_Tomorrow</author><pubDate>Mon, 25 Dec 2017 21:02:59 +0800</pubDate><description><![CDATA[输入为标量(Scalar)

f(xk+δ)&amp;nbsp;≈&amp;nbsp;f(xk)+f′(xk)δ+f′′(xk)δ22&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;δ+xk∈U˚(xk)f(xk+δ)&amp;nbsp;≈&amp;nbsp;f(xk)+f′(xk)δ+f″(xk)δ22&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;&amp;nbsp;δ+xk∈U˚]]></description><category></category></item><item><title><![CDATA[机器学习中评估算法的常用评价指标]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78896841</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78896841</guid><author>Recall_Tomorrow</author><pubDate>Mon, 25 Dec 2017 21:00:42 +0800</pubDate><description><![CDATA[评价指标

​   对于自然语言处理(NLP)、机器学习(ML)、信息检索(IR)等领域，评估(Evaluation)是一个必要的工作，而这其中所涉及到的评价指标一般包括：准确率(Accuracy)、精确率(Precision)、召回率(Recall)以及F1-measure

​   TP(True Positive)、TN(True Negative)、FP(False Positive)、F...]]></description><category></category></item><item><title><![CDATA[逻辑斯特回归(随机梯度上升算法)预测马疝病的死亡率问题]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78895256</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78895256</guid><author>Recall_Tomorrow</author><pubDate>Mon, 25 Dec 2017 17:54:08 +0800</pubDate><description><![CDATA[预测马疝病死亡率github代码随机梯度上升(下降)算法推导过程：使用的一些变量，类别标签向量yy，数据集样本矩阵XX，回归系数向量WW，观察值与真实值偏差向量ee，步长(学习率)α\alpha，PS:以上向量均为标准列向量。]]></description><category></category></item><item><title><![CDATA[隐含马尔科夫模型(HMM)原理及其实现]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78439395</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78439395</guid><author>Recall_Tomorrow</author><pubDate>Fri, 03 Nov 2017 20:22:43 +0800</pubDate><description><![CDATA[隐含马尔科夫模型
公式推导s1,s2,⋯,sn = Args1,s2,⋯,sn∈S Max P(s1,s2,⋯,sn|o1,o2,⋯,on)s_1,s_2,\cdots,s_n\ = \ \underbrace{Arg}_{s_1,s_2,\cdots,s_n\in S}\ Max\ P(s_1,s_2,\cdots,s_n|o_1,o_2,\cdots,o_n)           ⋯⋯\]]></description><category></category></item><item><title><![CDATA[三个字符串s1, s2, s3, 现要求你判断s3是否由s1和s2交错组成的]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78093853</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78093853</guid><author>Recall_Tomorrow</author><pubDate>Tue, 26 Sep 2017 12:07:27 +0800</pubDate><description><![CDATA[三个字符串s1, s2, s3, 现要求你判断s3是否由s1和s2交错组成的(如同归并排序两个有序数组合并为一个有序数组这样的交错)



package edu.ccnu;

import java.util.Scanner;

/**
 * 现有三个字符串s1, s2, s3, 现要求你判断s3是否由s1和s2交错组成的
 * @author MLee
 *
 */
public class ...]]></description><category></category></item><item><title><![CDATA[包含n个整数的数组，返回该数组和为最大的子数组]]></title><link>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78093821</link><guid>https://blog.youkuaiyun.com/Recall_Tomorrow/article/details/78093821</guid><author>Recall_Tomorrow</author><pubDate>Tue, 26 Sep 2017 12:01:53 +0800</pubDate><description><![CDATA[package edu.ccnu;

import java.util.Scanner;

import java.util.Scanner;

/**
 * 给定一个包含n个整数的数组，返回该数组和为最大的子数组(元素连续)
 * @author MLee
 *
 */
public class MaxSumOfSubarray {
    public static int getMaxSum...]]></description><category></category></item></channel></rss>