<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[吐槽NLP & ML]]></title><description><![CDATA[对NLP各种技术，以及机器学习的方法，进行全方位的吐槽]]></description><link>https://blog.youkuaiyun.com/marscrazy_90</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; marscrazy_90]]></copyright><item><title><![CDATA[对比学习文献综述]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/122820043</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/122820043</guid><author>marscrazy_90</author><pubDate>Tue, 08 Feb 2022 11:11:17 +0800</pubDate><description><![CDATA[对比学习综述 详细介绍了对比学习的起源，发展，从CPC到MAE，从图像到文本SimCSE，再到表格数据Tabular data，从入门到精通！！！]]></description><category></category></item><item><title><![CDATA[flask 四种wsgi方式对比（tornado，Gunicorn，Twisted，Gevent）]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/41943211</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/41943211</guid><author>marscrazy_90</author><pubDate>Mon, 15 Dec 2014 14:48:04 +0800</pubDate><description><![CDATA[使用工具：siege
代码结构：
hello.py
templates
|--hello.html
hello.py代码：
from flask import Flask, render_template
app = Flask(__name__)

@app.route('/')
@app.route('/')
def index(name=None):
    return ren]]></description><category></category></item><item><title><![CDATA[gnumpy numpy 性能对比]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/41941329</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/41941329</guid><author>marscrazy_90</author><pubDate>Mon, 15 Dec 2014 12:29:05 +0800</pubDate><description><![CDATA[要开始做实验，写代码了，想自己写一下工具，偏机器学习的，要用到矩阵运算比较多，于是想对numpy 的效率进行一下探索。
版本一
首先简单看一下矩阵乘法的结果，只用numpy：

import numpy as np
n = 4000
for i in range(10):
    a = np.random.uniform(low=0., high=1., size=(n, n)).asty]]></description><category></category></item><item><title><![CDATA[机器学习之路]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/26150655</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/26150655</guid><author>marscrazy_90</author><pubDate>Sun, 18 May 2014 17:51:46 +0800</pubDate><description><![CDATA[机器学习入门教材有许多，入门方式多种多样，我是半路出家，简单总结一下我的机器学习之路。2011年考入北邮计算机研究生学院，主要是NLP方向。之前没有接触过机器学习，本科也不是计算机专业，而是工商管理。


2011年研究生阶段，两节课对我影响很大：
第一节课是计算语言学，最开始了解和接触机器学习，是在上这节课的时候，老师推荐的教材《统计自然语言处理基础》。在老师的课上，有讲到HMM（隐马尔]]></description><category></category></item><item><title><![CDATA[python 参考手册 第五章]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25434343</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25434343</guid><author>marscrazy_90</author><pubDate>Sat, 10 May 2014 00:07:23 +0800</pubDate><description><![CDATA[循环加上数字索引]]></description><category></category></item><item><title><![CDATA[python参考手册 第四章]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25425305</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25425305</guid><author>marscrazy_90</author><pubDate>Fri, 09 May 2014 21:22:14 +0800</pubDate><description><![CDATA[首先是数学操作：
x//y  #截断除法]]></description><category></category></item><item><title><![CDATA[python参考手册 第三章]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25277757</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25277757</guid><author>marscrazy_90</author><pubDate>Thu, 08 May 2014 15:34:25 +0800</pubDate><description><![CDATA[比较两个对象身份与类型：


a  is  b #a和b是同一个]]></description><category></category></item><item><title><![CDATA[window 脚本]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25274737</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25274737</guid><author>marscrazy_90</author><pubDate>Thu, 08 May 2014 09:00:31 +0800</pubDate><description><![CDATA[昨天写脚本，有几条非常有用：]]></description><category></category></item><item><title><![CDATA[修炼 读书笔记 一]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25131857</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25131857</guid><author>marscrazy_90</author><pubDate>Tue, 06 May 2014 16:27:48 +0800</pubDate><description><![CDATA[系统设计七条原则：]]></description><category></category></item><item><title><![CDATA[python参考手册 第二章]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25128101</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25128101</guid><author>marscrazy_90</author><pubDate>Tue, 06 May 2014 15:02:54 +0800</pubDate><description><![CDATA[代码中空格和Tab混用的时候，可以使用]]></description><category></category></item><item><title><![CDATA[python参考手册 第一章]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/25125951</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/25125951</guid><author>marscrazy_90</author><pubDate>Tue, 06 May 2014 14:36:17 +0800</pubDate><description><![CDATA[if  key in map:
    p = map[key]
else:
    p=0.0

上面这段代码可以]]></description><category></category></item><item><title><![CDATA[Jython 和 Python 对比]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/24995427</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/24995427</guid><author>marscrazy_90</author><pubDate>Sun, 04 May 2014 17:55:24 +0800</pubDate><description><![CDATA[计算密集单线程 
0.0499999523163
计算密集多线程 
0.112999916077
IO密集单线程 
30.1949999332
IO密集多线程
3.13199996948]]></description><category></category></item><item><title><![CDATA[python unittest  测试框架]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/24993425</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/24993425</guid><author>marscrazy_90</author><pubDate>Sun, 04 May 2014 17:30:03 +0800</pubDate><description><![CDATA[以前使用Java 的时候有Junit单元测试，python据说]]></description><category></category></item><item><title><![CDATA[python __init__.py的作用  distutils 的使用]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/24981039</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/24981039</guid><author>marscrazy_90</author><pubDate>Sun, 04 May 2014 15:15:51 +0800</pubDate><description><![CDATA[怎么把自己写的python代码打包？]]></description><category></category></item><item><title><![CDATA[python c 扩展 ctype 以及 效率对比]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/14045683</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/14045683</guid><author>marscrazy_90</author><pubDate>Fri, 01 Nov 2013 22:34:35 +0800</pubDate><description><![CDATA[python开发效率高，第三方库资源丰富，而且可以用C/C++扩展。(基于python官方文档）
 
今天研究了用Ctype 来写python的扩展，用的是2.x版本的，刚开始还弄错了，看的是3.x版本的文档，浪费了不少时间。
首先将用C/C++ 编写python扩展吧；
我是在widows下，用的 vs2010 express，新建一个win32  DLL工程
然后在源码里面增加下面代]]></description><category></category></item><item><title><![CDATA[python 高级编程 四]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/13615273</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/13615273</guid><author>marscrazy_90</author><pubDate>Tue, 29 Oct 2013 23:30:52 +0800</pubDate><description><![CDATA[with 
 
with 常用应用场景：
打开文件
申请一个锁
 
可以实现with 操作的，实现了两个方法   __enter__ 和  __exit__
class TestWith(object):
    def __enter__(self):
        print "entering"
    
    def __exit__(self,exception]]></description><category></category></item><item><title><![CDATA[python 高级编程 三]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/13615065</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/13615065</guid><author>marscrazy_90</author><pubDate>Tue, 29 Oct 2013 23:18:42 +0800</pubDate><description><![CDATA[decorator
作用在某个函数上面，将其函数指针作为参数传入，然后添加一些操作，最后返回原来函数
两种方式，一种是无参的：
def Decorator(func):
    def new_func(*args,**kws):
        print "add some operations"
        func(*args,**kws)
    return new_]]></description><category></category></item><item><title><![CDATA[python 高级编程 二]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/13614479</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/13614479</guid><author>marscrazy_90</author><pubDate>Tue, 29 Oct 2013 23:02:46 +0800</pubDate><description><![CDATA[这次主要讲协同程序，和生成器表达式
 
协同程序：现在有twisted不知道是不是也是跟这个有关系。。
import multitask 
import time

def task1():
    for i in xrange(3):
        print "T1"
        yield i

def task2():
    for i in xrange]]></description><category></category></item><item><title><![CDATA[python 高级编程 一]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/13613333</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/13613333</guid><author>marscrazy_90</author><pubDate>Tue, 29 Oct 2013 22:18:47 +0800</pubDate><description><![CDATA[最近迷上了python，感觉python的语法简洁，功能丰富又强大。容易上手，学习曲线不像其他语言那么陡峭。
现在还有时间，可以将机器学习的算法和优化算法，用python来实现一下，不能一直都拿他们当黑盒来用。
试了试numpy 和scipy ，很给力！
 以前用python 写过一个多线程的爬虫，长时间抓取网页，结果出现了内存泄露，一直都找到原因，抓取个两三个小时就会因为内存爆掉而停止。]]></description><category></category></item><item><title><![CDATA[ＳＳＨ框架源码下载]]></title><link>https://blog.youkuaiyun.com/marscrazy_90/article/details/9257131</link><guid>https://blog.youkuaiyun.com/marscrazy_90/article/details/9257131</guid><author>marscrazy_90</author><pubDate>Sat, 06 Jul 2013 10:44:54 +0800</pubDate><description><![CDATA[用了一段时间的ＳＳＨ框架,想看看源码,深入研究一下的时候,找了很久没有找到.


好不容易找到了,又有段时间打不开,郁闷了一小会...


终于  ...
现在都是github了

spring 源码 下载地址 https://github.com/SpringSource/spring-framework
hibernate-orm 源码 下载地址 https://githu]]></description><category></category></item></channel></rss>