<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[黑洲非人lyf]]></title><description><![CDATA[共同学习进步, 联系邮箱:844460260@qq.com]]></description><link>https://blog.youkuaiyun.com/lyf52010</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; lyf52010]]></copyright><item><title><![CDATA[卷积神经网络（conv2d参数含义、卷积层、池化层）]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/119453685</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/119453685</guid><author>lyf52010</author><pubDate>Fri, 06 Aug 2021 14:08:48 +0800</pubDate><description><![CDATA[本文转载自

罗翌新：中科大数学博士，深度学习医学应用专家；
廖星宇：中科大硕士，计算机视觉专家，Face++资深工程师，《深度学习之Pytorch》作者；

的深度学习理论与实战（基于TensorFlow实现）

一、tf.nn.conv2d()卷积函数各参数解析













二、卷积例子









三、池化函数tf.nn.max_pool()函数解析











四、池化例子




...]]></description><category></category></item><item><title><![CDATA[深入理解java虚拟机-笔记]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/105405300</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/105405300</guid><author>lyf52010</author><pubDate>Thu, 09 Apr 2020 22:50:50 +0800</pubDate><description><![CDATA[java内存区域与内存溢出异常
java虚拟机自动内存管理机制，不用像C/C++为每一个new操作去写配对delete/free代码
java虚拟机在执行java程序的过程中   会把内存划分为若干个不同的数据区域
程序计数器:执行字节码指令 分支 循环 跳转。
java虚拟机的多线程是通过线程轮流切换并分配处理器执行时间的方式来实现的
java虚拟机栈 线程私有，存储局部变量
本地方法栈为nat...]]></description><category></category></item><item><title><![CDATA[插入排序和希尔排序]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/89499025</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/89499025</guid><author>lyf52010</author><pubDate>Wed, 24 Apr 2019 17:37:02 +0800</pubDate><description><![CDATA[可参考

https://github.com/hustcc/JS-Sorting-Algorithm

http://www.runoob.com/w3cnote/bubble-sort.html

插入排序

插入排序的代码实现虽然没有冒泡排序和选择排序那么简单粗暴，但它的原理应该是最容易理解的了，因为只要打过扑克牌的人都应该能够秒懂。插入排序是一种最简单直观的排序算法，它的工作原理是通过构建...]]></description><category></category></item><item><title><![CDATA[力扣刷算法题]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/89378428</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/89378428</guid><author>lyf52010</author><pubDate>Thu, 18 Apr 2019 14:39:52 +0800</pubDate><description><![CDATA[https://leetcode-cn.com/problems/majority-element/description/
]]></description><category></category></item><item><title><![CDATA[Jaccard和Levenshtein]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/89333598</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/89333598</guid><author>lyf52010</author><pubDate>Tue, 16 Apr 2019 15:04:09 +0800</pubDate><description><![CDATA[目录

Jaccard相似度

Levenshtein距离

Jaccard相似度

1、jaccard index又称为jaccard similarity coefficient用于比较有限样本集之间的相似性和差异性

定义：

给定两个集合A,B jaccard 系数定义为A与B交集的大小与并集大小的比值，jaccard值越大说明相似度越高

Levenshtein距离

拿idf还得自己去...]]></description><category></category></item><item><title><![CDATA[python知识点]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/89317189</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/89317189</guid><author>lyf52010</author><pubDate>Mon, 15 Apr 2019 18:06:48 +0800</pubDate><description><![CDATA[目录

python super详解

Python __call__函数

Self的定义

继承

python super详解

http://www.runoob.com/w3cnote/python-super-detail-intro.html


class A:
    def __init__(self):
        self.n = 2

    def add(self...]]></description><category></category></item><item><title><![CDATA[顺序表(线性表的顺序存储结构)及C语言实现]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/89020808</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/89020808</guid><author>lyf52010</author><pubDate>Thu, 04 Apr 2019 13:41:55 +0800</pubDate><description><![CDATA[#include &lt;stdio.h&gt;
#include &lt;string.h&gt;
#include &lt;stdlib.h&gt;
#define MAXSIZE 20
#define initSize 10
typedef int ElemType;
typedef struct{
	ElemType *data;
	int length;
}SeqList;

S...]]></description><category></category></item><item><title><![CDATA[SMOTE-类不平衡问题]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/88856364</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/88856364</guid><author>lyf52010</author><pubDate>Wed, 27 Mar 2019 21:48:55 +0800</pubDate><description><![CDATA[什么是类不平衡问题

  类不平衡（class-imbalance）是指在训练分类器中所使用的训练集的类别分布不均。比如说一个二分类问题，1000个训练样本，比较理想的情况是正类、负类样本的数量相差不多；而如果正类样本有995个、负类样本仅5个，就意味着存在类不平衡。

  在后文中，把样本数量过少的类别称为“少数类”。

   但实际上，数据集上的类不平衡到底有没有达到需要...]]></description><category></category></item><item><title><![CDATA[链表增加节点和反转java]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/88079129</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/88079129</guid><author>lyf52010</author><pubDate>Sat, 02 Mar 2019 17:59:32 +0800</pubDate><description><![CDATA[public class reverseLink {
	
	public static void main(String []args) {
		MyLinkList myLinkList = new reverseLink().new MyLinkList();

		myLinkList.addNode(1);
		myLinkList.addNode(2);
		myLinkList.ad...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战六(保存网络和加载网络)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/88064730</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/88064730</guid><author>lyf52010</author><pubDate>Fri, 01 Mar 2019 18:18:47 +0800</pubDate><description><![CDATA[保存网络


import tensorflow as tf
import numpy as np

# Save to file
# remember to define the same dtype and shape when restore
W = tf.Variable([[1,2,3],[3,4,5]], dtype=tf.float32, name='weights')
b = tf...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战五(卷积神经网络)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/88046701</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/88046701</guid><author>lyf52010</author><pubDate>Thu, 28 Feb 2019 22:56:52 +0800</pubDate><description><![CDATA[方法定义tf.nn.conv2d (input, filter, strides, padding, use_cudnn_on_gpu=None, data_format=None, name=None)

参数：**input : ** 输入的要做卷积的图片，要求为一个张量，shape为 [ batch, in_height, in_weight, in_channel ]，其中batch为图片...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战四(识别手写数字集mnist)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/88026964</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/88026964</guid><author>lyf52010</author><pubDate>Thu, 28 Feb 2019 14:22:46 +0800</pubDate><description><![CDATA[手写数字集mnist

任然是集合作为特征数*样本数，特征数代表了某层神经元数量

wx_plus_b = tf.nn.dropout(wx_plus_b,keep_prob)

dropout可以解决过拟合


def add_layer(inputs,in_size,out_size,activation_function=None):
    Weights = tf.Variable(tf...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战 拓展(Tensorflow常用函数)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87992599</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87992599</guid><author>lyf52010</author><pubDate>Wed, 27 Feb 2019 21:47:44 +0800</pubDate><description><![CDATA[tf.argmax        axis


y_pre = [[5,1,1,0,1],[2,3,-4,5,6]]
​
session.run(tf.arg_max(y_pre,1))
array([0, 4], dtype=int64)

y_pre
session.run(tf.arg_max(y_pre,0))
array([0, 1, 0, 1, 1], dtype=in...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战三(构造简单的神经网络)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87936204</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87936204</guid><author>lyf52010</author><pubDate>Wed, 27 Feb 2019 16:48:21 +0800</pubDate><description><![CDATA[1：tf.random_normal:

正态分布产生的随机值：常用的参数就是shape，和dtype了，但是也包括方差和均值；

参数（shape,stddev,mean,dtype)

2:tf.random_uniform 

默然是在0到1之间产生随机数：

但是也可以通过maxval指定上界，通过minval指定下界

np.newaxis 的实用

np.newaxis 在使用和功能上...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战一(变量与常量/placeholder)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87934891</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87934891</guid><author>lyf52010</author><pubDate>Tue, 26 Feb 2019 15:12:03 +0800</pubDate><description><![CDATA[变量与常量


import tensorflow as tf 

## 定义
#定义一个变量
var =tf.Variable(0,name=&quot;myvar&quot;)

#定义一个常量
con_var=tf.constant(1)

#定义一个加法
new_var=tf.add(var,con_var)

## 开始计算

#初始化，在初始化之前是变量是没有值的
init =tf.global_vari...]]></description><category></category></item><item><title><![CDATA[Tensorflow入门到实战二(求解线性回归)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87921760</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87921760</guid><author>lyf52010</author><pubDate>Mon, 25 Feb 2019 18:44:08 +0800</pubDate><description><![CDATA[求解线性回归

代码块如下


import tensorflow as tf
import numpy as np

x_data = np.random.rand(100).astype(np.float32)
y_data = 0.1*x_data +0.3
print(tf.random_uniform([1],-1,1))
weights = tf.Variable(tf.random_...]]></description><category></category></item><item><title><![CDATA[HMM与分词、词性标注、命名实体识别]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87862483</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87862483</guid><author>lyf52010</author><pubDate>Thu, 21 Feb 2019 15:58:04 +0800</pubDate><description><![CDATA[HMM自己的博客链接

以上为之前的HMM总结的有点抽象，这次来个实际的小项目来总结一下

参考自http://www.hankcs.com/nlp/hmm-and-segmentation-tagging-named-entity-recognition.html

目录

HMM描述

例子描述

求解最可能的天气

这个例子的Python代码：

NLP应用

小结

HMM（隐马尔可夫模型...]]></description><category></category></item><item><title><![CDATA[需要学习的网站/博客指南]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87811268</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87811268</guid><author>lyf52010</author><pubDate>Wed, 20 Feb 2019 18:00:30 +0800</pubDate><description><![CDATA[这个博客的博主的内容很好：http://www.hankcs.com/



tensorlow学习：https://www.jianshu.com/p/27a2fb320934

代码：https://github.com/zhaozhengcoder/Machine-Learning



tensorflow 学习csdn博客https://blog.youkuaiyun.com/xierhacke...]]></description><category></category></item><item><title><![CDATA[基于物品/用户/内容的推荐算法(ItemCF, UserCF, ContentCF)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87344691</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87344691</guid><author>lyf52010</author><pubDate>Fri, 15 Feb 2019 14:53:55 +0800</pubDate><description><![CDATA[课程链接:https://www.imooc.com/video/15790

代码链接: https://github.com/SkillyZ/java-spring/tree/master/skilly-hadoop

Hadoop的一些java访问接口编程步骤等：https://www.cnblogs.com/zhangyinhua/p/7678704.html#_lab2_1_1

 

...]]></description><category></category></item><item><title><![CDATA[Hadoop学习笔记(一)]]></title><link>https://blog.youkuaiyun.com/lyf52010/article/details/87258237</link><guid>https://blog.youkuaiyun.com/lyf52010/article/details/87258237</guid><author>lyf52010</author><pubDate>Thu, 14 Feb 2019 14:26:13 +0800</pubDate><description><![CDATA[HDFS架构图



HDFS:分布式文件系统

DataNode:存放文件或文件副本。块：最小64MB。





 

Map Reduce的四个阶段：

1、Split阶段（分片输入阶段）

2、Map阶段（需要编码）

3、Shuffle阶段

4、Reduce阶段（需要编码）



Word Count实例回顾：

1、Input（输入文件）

2、Split阶段（分片输入阶段），将数据...]]></description><category></category></item></channel></rss>