﻿<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:trackback="http://madskills.com/public/xml/rss/module/trackback/" xmlns:wfw="http://wellformedweb.org/CommentAPI/" xmlns:slash="http://purl.org/rss/1.0/modules/slash/"><channel><title>C++博客-蔡东赟-随笔分类-关于人工智能的yy</title><link>http://www.cppblog.com/cdy20/category/20044.html</link><description>qq：286402937  
weibo： http://weibo.com/caidongyun
架构-开源交流 167813458
lomox群：41830909
更多资料：微刊  http://t.cn/zT53Ryw  
</description><language>zh-cn</language><lastBuildDate>Wed, 14 Oct 2015 10:44:28 GMT</lastBuildDate><pubDate>Wed, 14 Oct 2015 10:44:28 GMT</pubDate><ttl>60</ttl><item><title>Deep Learning (Spark, Caffe, GPU) </title><link>http://www.cppblog.com/cdy20/archive/2015/10/14/212013.html</link><dc:creator>蔡东赟</dc:creator><author>蔡东赟</author><pubDate>Wed, 14 Oct 2015 09:25:00 GMT</pubDate><guid>http://www.cppblog.com/cdy20/archive/2015/10/14/212013.html</guid><wfw:comment>http://www.cppblog.com/cdy20/comments/212013.html</wfw:comment><comments>http://www.cppblog.com/cdy20/archive/2015/10/14/212013.html#Feedback</comments><slash:comments>0</slash:comments><wfw:commentRss>http://www.cppblog.com/cdy20/comments/commentRss/212013.html</wfw:commentRss><trackback:ping>http://www.cppblog.com/cdy20/services/trackbacks/212013.html</trackback:ping><description><![CDATA[<h1 style="widows: auto; color: #4a4c4e; font-family: Lato, sans-serif; font-size: 34px; font-weight: 300; background-color: #ffffff;"></h1><h1 style="color: #4a4c4e; font-family: Lato, sans-serif; font-size: 34px; font-weight: 300; background-color: #ffffff;"><span style="font-size: 8pt;">from http://docs.continuum.io/anaconda-cluster/examples/spark-caffe</span><br /><br />Deep Learning (Spark, Caffe, GPU)</h1><h1><div style="display: inline-block;"></div></h1><h2 style="color: #737272; font-family: inherit; font-size: 20px; font-weight: 400;">Description</h2><h1><div style="display: inline-block;"><div style="font-weight: normal; widows: auto; color: #333333; font-family: Lato, sans-serif; font-size: 13px; background-color: #ffffff;"><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">To demonstrate the capability of running a distributed job in PySpark using a GPU, this example uses a neural network library,<span style="background-color: inherit;">&nbsp;</span><a href="http://caffe.berkeleyvision.org/" style="background-color: inherit; cursor: pointer; color: #0088cc;">Caffe</a>. Below is a trivial example of using Caffe on a Spark cluster; although this is redundant, it demonstrates the capability of training neural networks with GPUs.</p><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">For this example, we recommend the use of the AMI<span style="background-color: inherit;">&nbsp;</span><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; background-color: inherit;"><span style="background-color: inherit;">ami-2cbf3e44</span></code><span style="background-color: inherit;">&nbsp;</span>and the instance type<span style="background-color: inherit;">&nbsp;</span><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; background-color: inherit;"><span style="background-color: inherit;">g2.2xlarge</span></code>. An example profile (to be placed in<span style="background-color: inherit;">&nbsp;</span><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; background-color: inherit;"><span style="background-color: inherit;">~/.acluster/profiles.d/gpu_profile.yaml</span></code>) is shown below:</p><div style="background-color: #eeffcc;"><pre style="font-family: Menlo, Monaco, Consolas, 'Courier New', monospace; font-size: 12.025px; background-color: #f5f5f5;">name: gpu_profile
node_id: ami-2cbf3e44 # Ubuntu 14.04 - IS HVM - Cuda 6.5
user: ubuntu
node_type: g2.2xlarge
num_nodes: 3
provider: aws
plugins:
  - spark-yarn
  - notebook
</pre></div></div></div></h1><h2 style="color: #737272; font-family: inherit; font-size: 20px; font-weight: 400;">Download</h2><h1><div style="display: inline-block;"><div style="font-weight: normal; widows: auto; color: #333333; font-family: Lato, sans-serif; font-size: 13px; background-color: #ffffff;"><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">To execute this example, download the:<span style="background-color: inherit;">&nbsp;</span><a href="http://docs.continuum.io/_downloads/spark-caffe.py" style="background-color: inherit; cursor: pointer; color: #0088cc;"><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; font-weight: bold; background-color: transparent;"><span style="background-color: inherit;">spark-caffe.py</span><span style="background-color: inherit;">&nbsp;</span><span style="background-color: inherit;">example</span><span style="background-color: inherit;">&nbsp;</span><span style="background-color: inherit;">script</span></code></a><span style="background-color: inherit;">&nbsp;</span>or<span style="background-color: inherit;">&nbsp;</span><a href="http://docs.continuum.io/_downloads/spark-caffe.ipynb" style="background-color: inherit; cursor: pointer; color: #0088cc;"><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; font-weight: bold; background-color: transparent;"><span style="background-color: inherit;">spark-caffe.ipynb</span><span style="background-color: inherit;">example</span><span style="background-color: inherit;">&nbsp;</span><span style="background-color: inherit;">notebook</span></code></a>.</p></div></div></h1><h2 style="color: #737272; font-family: inherit; font-size: 20px; font-weight: 400;">Installation</h2><h1><div style="display: inline-block;"><div style="font-weight: normal; widows: auto; color: #333333; font-family: Lato, sans-serif; font-size: 13px; background-color: #ffffff;"><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">The Spark + YARN plugin can be installed on the cluster using the following command:</p><div style="background-color: #eeffcc;"><pre style="font-family: Menlo, Monaco, Consolas, 'Courier New', monospace; font-size: 12.025px; background-color: #f5f5f5;"><span style="background-color: inherit; color: #bb60d5;">$ </span>acluster install spark-yarn
</pre></div><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">Once the Spark + YARN plugin is installed, you can view the YARN UI in your browser using the following command:</p><div style="background-color: #eeffcc;"><pre style="font-family: Menlo, Monaco, Consolas, 'Courier New', monospace; font-size: 12.025px; background-color: #f5f5f5;"><span style="background-color: inherit; color: #bb60d5;">$ </span>acluster open yarn
</pre></div></div></div></h1><h2 style="color: #737272; font-family: inherit; font-size: 20px; font-weight: 400;">Dependencies</h2><h1><div style="display: inline-block;"><div style="font-weight: normal; widows: auto; color: #333333; font-family: Lato, sans-serif; font-size: 13px; background-color: #ffffff;"><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">First, we need to bootstrap Caffe and its dependencies on all of the nodes. We provide a bash script that will install Caffe from source:<span style="background-color: inherit;">&nbsp;</span><a href="http://docs.continuum.io/_downloads/bootstrap-caffe.sh" style="background-color: inherit; cursor: pointer; color: #0088cc;"><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; font-weight: bold; background-color: transparent;">bootstrap-caffe.sh</code></a>. The following command can be used to upload the<span style="background-color: inherit;">&nbsp;</span><code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; background-color: inherit;"><span style="background-color: inherit;">bootstrap-caffe.sh</span></code><span style="background-color: inherit;">&nbsp;</span>script to all of the nodes and execute it in parallel:</p><div style="background-color: #eeffcc;"><pre style="font-family: Menlo, Monaco, Consolas, 'Courier New', monospace; font-size: 12.025px; background-color: #f5f5f5;">$ acluster submit bootstrap-caffe.sh --all
</pre></div><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">After a few minues, Caffe and its dependencies will be installed on the cluster nodes and the job can be started.</p></div></div></h1><h2 style="color: #737272; font-family: inherit; font-size: 20px; font-weight: 400;">Running the Job</h2><h1><div style="font-weight: normal; widows: auto; color: #333333; font-family: Lato, sans-serif; font-size: 13px; background-color: #ffffff;"><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">Here is the complete script to run the Spark + GPU with Caffe example in PySpark:</p><div style="background-color: #eeffcc;"><pre style="font-family: Menlo, Monaco, Consolas, 'Courier New', monospace; font-size: 12.025px; background-color: #f5f5f5;"><span style="background-color: inherit; color: #408090; font-style: italic;"># spark-caffe.py</span> <span style="background-color: inherit; color: #007020; font-weight: bold;">from</span> <span style="background-color: inherit; color: #0e84b5; font-weight: bold;">pyspark</span> <span style="background-color: inherit; color: #007020; font-weight: bold;">import</span> <span style="background-color: inherit;">SparkConf</span> <span style="background-color: inherit; color: #007020; font-weight: bold;">from</span> <span style="background-color: inherit; color: #0e84b5; font-weight: bold;">pyspark</span> <span style="background-color: inherit; color: #007020; font-weight: bold;">import</span> <span style="background-color: inherit;">SparkContext</span>  <span style="background-color: inherit;">conf</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">SparkConf</span><span style="background-color: inherit;">()</span> <span style="background-color: inherit;">conf</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">setMaster</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #4070a0;">'yarn-client'</span><span style="background-color: inherit;">)</span> <span style="background-color: inherit;">conf</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">setAppName</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #4070a0;">'spark-caffe'</span><span style="background-color: inherit;">)</span> <span style="background-color: inherit;">sc</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">SparkContext</span><span style="background-color: inherit;">(</span><span style="background-color: inherit;">conf</span><span style="background-color: inherit; color: #666666;">=</span><span style="background-color: inherit;">conf</span><span style="background-color: inherit;">)</span>   <span style="background-color: inherit; color: #007020; font-weight: bold;">def</span> <span style="background-color: inherit; color: #06287e;">noop</span><span style="background-color: inherit;">(</span><span style="background-color: inherit;">x</span><span style="background-color: inherit;">):</span>     <span style="background-color: inherit; color: #007020; font-weight: bold;">import</span> <span style="background-color: inherit; color: #0e84b5; font-weight: bold;">socket</span>     <span style="background-color: inherit; color: #007020; font-weight: bold;">return</span> <span style="background-color: inherit;">socket</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">gethostname</span><span style="background-color: inherit;">()</span>  <span style="background-color: inherit;">rdd</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">sc</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">parallelize</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #007020;">range</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #208050;">2</span><span style="background-color: inherit;">),</span> <span style="background-color: inherit; color: #208050;">2</span><span style="background-color: inherit;">)</span> <span style="background-color: inherit;">hosts</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">rdd</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">map</span><span style="background-color: inherit;">(</span><span style="background-color: inherit;">noop</span><span style="background-color: inherit;">)</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">distinct</span><span style="background-color: inherit;">()</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">collect</span><span style="background-color: inherit;">()</span> <span style="background-color: inherit; color: #007020; font-weight: bold;">print</span> <span style="background-color: inherit;">hosts</span>   <span style="background-color: inherit; color: #007020; font-weight: bold;">def</span> <span style="background-color: inherit; color: #06287e;">caffe_process</span><span style="background-color: inherit;">(</span><span style="background-color: inherit;">x</span><span style="background-color: inherit;">):</span>     <span style="background-color: inherit; color: #007020; font-weight: bold;">import</span> <span style="background-color: inherit; color: #0e84b5; font-weight: bold;">os</span>     <span style="background-color: inherit;">os</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">environ</span><span style="background-color: inherit;">[</span><span style="background-color: inherit; color: #4070a0;">'PATH'</span><span style="background-color: inherit;">]</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit; color: #4070a0;">'/usr/local/cuda/bin'</span> <span style="background-color: inherit; color: #666666;">+</span> <span style="background-color: inherit; color: #4070a0;">':'</span> <span style="background-color: inherit; color: #666666;">+</span> <span style="background-color: inherit;">os</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">environ</span><span style="background-color: inherit;">[</span><span style="background-color: inherit; color: #4070a0;">'PATH'</span><span style="background-color: inherit;">]</span>     <span style="background-color: inherit;">os</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">environ</span><span style="background-color: inherit;">[</span><span style="background-color: inherit; color: #4070a0;">'LD_LIBRARY_PATH'</span><span style="background-color: inherit;">]</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit; color: #4070a0;">'/usr/local/cuda/lib64:/home/ubuntu/pombredanne-https-gitorious.org-mdb-mdb.git-9cc04f604f80/libraries/liblmdb'</span>     <span style="background-color: inherit; color: #007020; font-weight: bold;">import</span> <span style="background-color: inherit; color: #0e84b5; font-weight: bold;">subprocess</span>     <span style="background-color: inherit;">proc</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">subprocess</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">Popen</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #4070a0;">'cd /home/ubuntu/caffe &amp;&amp; bash ./examples/mnist/train_lenet.sh'</span><span style="background-color: inherit;">,</span> <span style="background-color: inherit;">shell</span><span style="background-color: inherit; color: #666666;">=</span><span style="background-color: inherit; color: #007020;">True</span><span style="background-color: inherit;">,</span> <span style="background-color: inherit;">stdout</span><span style="background-color: inherit; color: #666666;">=</span><span style="background-color: inherit;">subprocess</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">PIPE</span><span style="background-color: inherit;">,</span> <span style="background-color: inherit;">stderr</span><span style="background-color: inherit; color: #666666;">=</span><span style="background-color: inherit;">subprocess</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">PIPE</span><span style="background-color: inherit;">)</span>     <span style="background-color: inherit;">out</span><span style="background-color: inherit;">,</span> <span style="background-color: inherit;">err</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">proc</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">communicate</span><span style="background-color: inherit;">()</span>     <span style="background-color: inherit; color: #007020; font-weight: bold;">return</span> <span style="background-color: inherit;">proc</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">returncode</span><span style="background-color: inherit;">,</span> <span style="background-color: inherit;">out</span><span style="background-color: inherit;">,</span> <span style="background-color: inherit;">err</span>  <span style="background-color: inherit;">rdd</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">sc</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">parallelize</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #007020;">range</span><span style="background-color: inherit;">(</span><span style="background-color: inherit; color: #208050;">2</span><span style="background-color: inherit;">),</span> <span style="background-color: inherit; color: #208050;">2</span><span style="background-color: inherit;">)</span> <span style="background-color: inherit;">ret</span> <span style="background-color: inherit; color: #666666;">=</span> <span style="background-color: inherit;">rdd</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">map</span><span style="background-color: inherit;">(</span><span style="background-color: inherit;">caffe_process</span><span style="background-color: inherit;">)</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">distinct</span><span style="background-color: inherit;">()</span><span style="background-color: inherit; color: #666666;">.</span><span style="background-color: inherit;">collect</span><span style="background-color: inherit;">()</span> <span style="background-color: inherit; color: #007020; font-weight: bold;">print</span> <span style="background-color: inherit;">ret</span> </pre></div><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">You can submit the script to the Spark cluster using the&nbsp;<code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; background-color: inherit;"><span style="background-color: inherit;">submit</span></code>&nbsp;command.</p><div style="background-color: #eeffcc;"><pre style="font-family: Menlo, Monaco, Consolas, 'Courier New', monospace; font-size: 12.025px; background-color: #f5f5f5;">$ acluster submit spark-caffe.py </pre></div><p style="margin-right: 0px; margin-left: 0px; font-family: 'Helvetica Neue', Helvetica, Arial, sans-serif; background-color: inherit;">After the script completes, the trained Caffe model can be found at<code style="font-family: Monaco, Consolas, Courier, 'Lucida Console', monospace; background-color: inherit;"><span style="background-color: inherit;">/home/ubuntu/caffe/examples/mnist/lenet_iter_10000.caffemodel</span></code>&nbsp;on all of the compute nodes.</p><div></div></div></h1><img src ="http://www.cppblog.com/cdy20/aggbug/212013.html" width = "1" height = "1" /><br><br><div align=right><a style="text-decoration:none;" href="http://www.cppblog.com/cdy20/" target="_blank">蔡东赟</a> 2015-10-14 17:25 <a href="http://www.cppblog.com/cdy20/archive/2015/10/14/212013.html#Feedback" target="_blank" style="text-decoration:none;">发表评论</a></div>]]></description></item><item><title>状态机-》智能 设想1</title><link>http://www.cppblog.com/cdy20/archive/2013/01/22/197464.html</link><dc:creator>蔡东赟</dc:creator><author>蔡东赟</author><pubDate>Tue, 22 Jan 2013 05:11:00 GMT</pubDate><guid>http://www.cppblog.com/cdy20/archive/2013/01/22/197464.html</guid><wfw:comment>http://www.cppblog.com/cdy20/comments/197464.html</wfw:comment><comments>http://www.cppblog.com/cdy20/archive/2013/01/22/197464.html#Feedback</comments><slash:comments>0</slash:comments><wfw:commentRss>http://www.cppblog.com/cdy20/comments/commentRss/197464.html</wfw:commentRss><trackback:ping>http://www.cppblog.com/cdy20/services/trackbacks/197464.html</trackback:ping><description><![CDATA[<div style="background-color:#eeeeee;font-size:13px;border:1px solid #CCCCCC;padding-right: 5px;padding-bottom: 4px;padding-left: 4px;padding-top: 4px;width: 98%;word-break:break-all"><div style="border: 1px solid #cccccc; padding: 4px 5px 4px 4px; width: 98%; word-break: break-all; "><!--<br /><br />Code highlighting produced by Actipro CodeHighlighter (freeware)<br />http://www.CodeHighlighter.com/<br /><br />-->昨晚忙完无聊的事情，想到了状态机，和智能。<br />关于低等智能进化的一个体系结构。可能有点凌乱，先记下来。<br /><br /><br />我们用类来描述<br /><br />Class<br />{<br />数据：<br />　　状态规则。State&nbsp;&nbsp;map&nbsp;or&nbsp;tree<br />　　状态机机制保存。Each&nbsp;State&nbsp;（data，&nbsp;rule，&nbsp;function）<br />　　<br />操作：<br />　　<br />　　SateAction(环境变量)<br />　　{<br />Switch（state）<br />{<br />Case&nbsp;state1：<br />State1_function();<br />Break:<br />Case&nbsp;state2：<br />State2_function();<br />Break:<br />　　Defualt:<br />}<br />　　<br />　　<span style="color: #008000; ">//</span><span style="color: #008000; ">记录&nbsp;行为状态<br />　　</span><span style="color: #008000; ">//</span><span style="color: #008000; ">进化操作</span><span style="color: #008000; "><br /></span>　　}<br />　　<br />　　<br />　　<span style="color: #008000; ">//</span><span style="color: #008000; ">输入为：外部视野&nbsp;内或者是&nbsp;感知范围&nbsp;环境&nbsp;驱动整个对象&nbsp;</span><span style="color: #008000; "><br /></span>　　Update()<br />　　{<br />　　}<br />　　<br />进化操作：<br />　　<span style="color: #008000; ">//</span><span style="color: #008000; ">不断吞噬其他对象的&nbsp;状态机接入进来，连接点为&nbsp;共同的外部环境变量</span><span style="color: #008000; "><br /></span>　　Void&nbsp;Eat(other);<br />　　<br />　　<span style="color: #008000; ">//</span><span style="color: #008000; ">自我角色&nbsp;和环境&nbsp;评估</span><span style="color: #008000; "><br /></span>　　Void&nbsp;Evolution();&nbsp;<span style="color: #008000; ">//</span><span style="color: #008000; ">自我扩展，各种进化&nbsp;方式<br />　　<br />　　</span><span style="color: #008000; ">//</span><span style="color: #008000; ">状态机自我扩展操作</span><span style="color: #008000; "><br /></span>　　Void&nbsp;EvolutionGo();<br />　　<br />}<br /><br /></div></div><img src ="http://www.cppblog.com/cdy20/aggbug/197464.html" width = "1" height = "1" /><br><br><div align=right><a style="text-decoration:none;" href="http://www.cppblog.com/cdy20/" target="_blank">蔡东赟</a> 2013-01-22 13:11 <a href="http://www.cppblog.com/cdy20/archive/2013/01/22/197464.html#Feedback" target="_blank" style="text-decoration:none;">发表评论</a></div>]]></description></item><item><title>关于智能 人与人之间 语言反馈</title><link>http://www.cppblog.com/cdy20/archive/2012/12/26/196676.html</link><dc:creator>蔡东赟</dc:creator><author>蔡东赟</author><pubDate>Wed, 26 Dec 2012 06:25:00 GMT</pubDate><guid>http://www.cppblog.com/cdy20/archive/2012/12/26/196676.html</guid><wfw:comment>http://www.cppblog.com/cdy20/comments/196676.html</wfw:comment><comments>http://www.cppblog.com/cdy20/archive/2012/12/26/196676.html#Feedback</comments><slash:comments>0</slash:comments><wfw:commentRss>http://www.cppblog.com/cdy20/comments/commentRss/196676.html</wfw:commentRss><trackback:ping>http://www.cppblog.com/cdy20/services/trackbacks/196676.html</trackback:ping><description><![CDATA[<div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">关于智能 人与人之间 语言反馈</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">早上公交车上面，收音机听到在讲这个词</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; "><br />信息：</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; "><span style="font-size: 12pt; ">&#8220;你这种人&#8221; <br /></span>&nbsp; &nbsp;<br /><span style="font-size: 12pt; ">假设给他一个语气</span></div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; "></div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">联想到的关于智能之间的交互。</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; "></div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">语言应激性</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">-&gt;历史背景 产生的定义 -》定义推出出来的贬义</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">-&gt;贬义产生的历史记忆 推导-》身体动物性变化</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">-&gt;化学物质变化-&gt;抑制刺激激励 沟通排斥，身体器官活动等 包括语言反馈 大脑反馈 其他内脏反馈</div><div style="font-family: 微软雅黑; line-height: normal; text-align: -webkit-auto; ">-&gt;记忆存储，历史库</div><img src ="http://www.cppblog.com/cdy20/aggbug/196676.html" width = "1" height = "1" /><br><br><div align=right><a style="text-decoration:none;" href="http://www.cppblog.com/cdy20/" target="_blank">蔡东赟</a> 2012-12-26 14:25 <a href="http://www.cppblog.com/cdy20/archive/2012/12/26/196676.html#Feedback" target="_blank" style="text-decoration:none;">发表评论</a></div>]]></description></item><item><title>机器学习相关数据库 备份</title><link>http://www.cppblog.com/cdy20/archive/2012/10/10/193134.html</link><dc:creator>蔡东赟</dc:creator><author>蔡东赟</author><pubDate>Wed, 10 Oct 2012 11:53:00 GMT</pubDate><guid>http://www.cppblog.com/cdy20/archive/2012/10/10/193134.html</guid><wfw:comment>http://www.cppblog.com/cdy20/comments/193134.html</wfw:comment><comments>http://www.cppblog.com/cdy20/archive/2012/10/10/193134.html#Feedback</comments><slash:comments>1</slash:comments><wfw:commentRss>http://www.cppblog.com/cdy20/comments/commentRss/193134.html</wfw:commentRss><trackback:ping>http://www.cppblog.com/cdy20/services/trackbacks/193134.html</trackback:ping><description><![CDATA[<br /><br /><br /><h3>资料库</h3><ul style="font-family: arial, verdana, tahoma, geneva; line-height: normal; "><li style="padding: 2px; "><strong><a href="http://www.sigkdd.org/kddcup/index.php" style="color: #551a8b; text-decoration: none; ">KDD杯的中心</a>，所有的数据，任务和结果。</strong></li><li style="padding: 2px; "><a href="http://kdd.ics.uci.edu/" style="color: #551a8b; text-decoration: none; "><strong>UCI</strong></a>机器学习和知识发现研究中使用的大型数据集<a href="http://kdd.ics.uci.edu/" style="color: #551a8b; text-decoration: none; "><strong>KDD数据库存储库</strong></a>。</li><li style="padding: 2px; "><a href="http://archive.ics.uci.edu/ml/" style="color: #551a8b; text-decoration: none; ">UCI机器学习数据库</a>。</li><li style="padding: 2px; "><a href="http://aws.amazon.com/publicdatasets/" style="color: #551a8b; text-decoration: none; ">AWS（亚马逊网络服务）公共数据集</a>，提供了一个集中的资料库，可以无缝集成到基于AWS的云应用程序的公共数据集。</li><li style="padding: 2px; "><a href="http://www.jcheminf.com/content/pdf/1758-2946-1-21.pdf" style="color: #551a8b; text-decoration: none; ">生物测定数据</a>，在&nbsp;<em>虚拟筛选，生物测定数据</em>，对化学信息学，J.由阿曼达Schierz的，有21个生物测定数据集（有效/无效的化合物）可供下载。</li><li style="padding: 2px; "><a href="http://www.data.gc.ca/" style="color: #551a8b; text-decoration: none; ">加拿大开放数据</a>，许多政府和地理空间数据集的试点项目。</li><li style="padding: 2px; "><a href="http://www.causality.inf.ethz.ch/repository.php" style="color: #551a8b; text-decoration: none; ">因果关系工作平台</a>的数据存储库。</li><li style="padding: 2px; "><a href="http://shop.oreilly.com/product/0636920018254.do" style="color: #551a8b; text-decoration: none; ">数据源手册</a>，指南公开数据，由皮特&#183;沃登，奥莱利（2011年1月）。</li><li style="padding: 2px; "><a href="http://data.gov.uk/" style="color: #551a8b; text-decoration: none; ">Data.gov.uk</a>，英国（<a href="http://data.london.gov.uk/" style="color: #551a8b; text-decoration: none; ">伦敦数据存储</a>的公开数据&nbsp;）。</li><li style="padding: 2px; "><a href="http://www.data.gov/education" style="color: #551a8b; text-decoration: none; ">Data.gov /教育</a>，中央指导教育数据资源，包括高价值的数据集，数据可视化工具，课堂上的资源，创建开放数据和更多的应用程序。</li><li style="padding: 2px; "><a href="http://datamarket.com/" style="color: #551a8b; text-decoration: none; ">DataMarket</a>，可视化世界的经济，社会，自然，与行业，与100万次系列由联合国，世界银行，欧盟统计局和其他重要的数据提供者。</li><li style="padding: 2px; "><a href="http://datamob.org/" style="color: #551a8b; text-decoration: none; ">Datamob</a>，公共数据善加利用。</li><li style="padding: 2px; "><a href="http://datasf.org/" style="color: #551a8b; text-decoration: none; ">DataSF.org</a>，从旧金山市与县，CA的数据资料交换所。</li><li style="padding: 2px; "><a href="http://dataferrett.census.gov/" style="color: #551a8b; text-decoration: none; ">DataFerrett</a>，数据挖掘工具，访问和操纵TheDataWeb，许多美国上线的电子政务数据集的集合。</li><li style="padding: 2px; "><a href="http://www.cs.toronto.edu/~delve" style="color: #551a8b; text-decoration: none; ">钻研</a>，评估学习有效的实验数据</li><li style="padding: 2px; "><a href="http://inforumweb.umd.edu/econdata/econdata.html" style="color: #551a8b; text-decoration: none; ">EconData</a>，成千上万的经济时间序列，产生一些美国政府机构。</li><li style="padding: 2px; "><a href="http://www.cs.cmu.edu/~enron/" style="color: #551a8b; text-decoration: none; ">安然邮件数据集</a>，约150个用户的数据，主要是安然公司的高级管理人员。</li><li style="padding: 2px; "><img src="http://www.kdnuggets.com/images/newr.gif" alt="新" height="12" width="25" />&nbsp;<a href="http://data.europeana.eu/" style="color: #551a8b; text-decoration: none; ">Europeana的数据</a>，包含2000万的文本，图像，视频和声音Europeana的-欧洲文化遗产内容的信任和全面的资源收集的开放的元数据。</li><li style="padding: 2px; "><a href="http://www.fedstats.gov/" style="color: #551a8b; text-decoration: none; ">FEDSTATS</a>，美国的统计数据，更全面的信息来源</li><li style="padding: 2px; "><a href="http://fimi.cs.helsinki.fi/" style="color: #551a8b; text-decoration: none; ">FIMI频繁项集挖掘</a>，实现和数据集<a href="http://fimi.cs.helsinki.fi/" style="color: #551a8b; text-decoration: none; ">库</a>。</li><li style="padding: 2px; "><a href="http://fisher.osu.edu/fin/fdf/osudata.htm" style="color: #551a8b; text-decoration: none; ">财务数据搜索在OSU</a>，大的目录的财务数据集。</li><li style="padding: 2px; "><a href="http://www.ncbi.nlm.nih.gov/geo/" style="color: #551a8b; text-decoration: none; ">全球环境展望（GEO基因表达OMNIBUS）&nbsp;</a>，基因表达/分子的丰富存储库支持MIAME兼容的数据提交，并策划，网络资源的基因表达数据的浏览，查询和检索。</li><li style="padding: 2px; "><a href="http://geodacenter.asu.edu/datalist/" style="color: #551a8b; text-decoration: none; ">GeoDa中心</a>，地理空间数据。</li><li style="padding: 2px; "><a href="http://ngrams.googlelabs.com/datasets" style="color: #551a8b; text-decoration: none; ">的谷歌ngrams数据集</a>，从数以百万计的谷歌扫描图书的文字。</li><li style="padding: 2px; "><a href="http://www.grainmarketresearch.com/" style="color: #551a8b; text-decoration: none; ">粮食市场调研</a>，财务数据，包括股票，期货等。</li><li style="padding: 2px; "><a href="https://bitly.com/bundles/hmason/1" style="color: #551a8b; text-decoration: none; ">希拉里&#183;梅森研究质量的大数据集</a>的集合-许多文本和图像数据集。</li><li style="padding: 2px; "><a href="http://www.icwsm.org/2009/data/" style="color: #551a8b; text-decoration: none; ">ICWSM-2009年的数据集</a>包含4400万8月1日和2008年10月1日的博客文章。</li><li style="padding: 2px; "><a href="http://infochimps.org/" style="color: #551a8b; text-decoration: none; ">Infochimps</a>，公开目录和市场数据。您可以共享，出售，牧师和下载任何东西，一切有关的数据。</li><li style="padding: 2px; "><a href="http://www.investorlinks.com/" style="color: #551a8b; text-decoration: none; ">投资者的链接</a>，包括财务数据</li><li style="padding: 2px; "><a href="http://kevinchai.net/datasets/" style="color: #551a8b; text-decoration: none; ">凯文湾仔数据集列表</a>，文本，SNA，和其他领域。</li><li style="padding: 2px; "><a href="http://konect.uni-koblenz.de/" style="color: #551a8b; text-decoration: none; ">KONECT</a>，科布伦茨网络的收集，与所有类型的大型网络数据集，以执行网络挖掘领域的研究。</li><li style="padding: 2px; "><a href="http://www.w3.org/wiki/SweoIG/TaskForces/CommunityProjects/LinkingOpenData" style="color: #551a8b; text-decoration: none; ">关联开放数据</a>项目，数据免费提供给大家。</li><li style="padding: 2px; "><a href="http://labrosa.ee.columbia.edu/millionsong/" style="color: #551a8b; text-decoration: none; ">万首歌曲数据集</a></li><li style="padding: 2px; "><a href="http://www-genome.wi.mit.edu/cgi-bin/cancer/datasets.cgi" style="color: #551a8b; text-decoration: none; ">麻省理工学院癌症基因组基因表达数据和出版物</a>，从麻省理工学院怀特黑德基因组研究中心。</li><li style="padding: 2px; "><a href="http://mldata.org/" style="color: #551a8b; text-decoration: none; ">ML数据</a>，数据存储库的，欧盟Pascal2网络。</li><li style="padding: 2px; "><a href="https://data.nasdaq.com/" style="color: #551a8b; text-decoration: none; ">纳斯达克Data Store数据存储系统</a>，提供市场数据的访问。</li><li style="padding: 2px; "><a href="http://www.archive-it.org/" style="color: #551a8b; text-decoration: none; ">国民政府统计网站</a>，数据，报告，统计年鉴，新闻发布，以及来自约70个网站，其中包括来自非洲，欧洲，亚洲和拉丁美洲的国家。</li><li style="padding: 2px; "><a href="http://nssdc.gsfc.nasa.gov/" style="color: #551a8b; text-decoration: none; ">国家空间科学数据中心</a>（NSSDC），美国航空航天局的数据集从行星探测，空间和太阳物理，生命科学，天体物理学，和更多。</li><li style="padding: 2px; "><a href="http://opendata.socrata.com/" style="color: #551a8b; text-decoration: none; ">OpenData Socrata</a>，访问超过10,000的数据集，包括商业，教育，政府和乐趣。</li><li style="padding: 2px; "><a href="http://www.delicious.com/pskomoroch/dataset" style="color: #551a8b; text-decoration: none; ">彼得Skomoroch数据集书签</a></li><li style="padding: 2px; "><a href="http://www.pubgene.org/" style="color: #551a8b; text-decoration: none; ">PubGene（TM）基因数据库和工具</a>，基因组学相关的出版物数据库</li><li style="padding: 2px; "><a href="http://www.econ.yale.edu/~shiller/data.htm" style="color: #551a8b; text-decoration: none; ">罗伯特&#183;席勒</a>住房，股市，并从他的著 &#8203;&#8203;作<em>&#8220;非理性繁荣的</em><a href="http://www.econ.yale.edu/~shiller/data.htm" style="color: #551a8b; text-decoration: none; ">数据</a>。</li><li style="padding: 2px; "><a href="http://genome-www5.stanford.edu/MicroArray/SMD/" style="color: #551a8b; text-decoration: none; ">SMD：斯坦福微阵列数据库</a>，从微阵列实验的原料和标准化的数据存储。</li><li style="padding: 2px; "><a href="http://www.nd.edu/~oss/Data/data.html" style="color: #551a8b; text-decoration: none; ">SourceForge.net研究资料</a>，包括历史和状态统计，约10万的项目和超过100万注册用户的活动，在项目管理网站。</li><li style="padding: 2px; "><a href="http://lib.stat.cmu.edu/datasets/" style="color: #551a8b; text-decoration: none; ">StatLib</a>，债务工具中央结算系统的数据集存档。</li><li style="padding: 2px; "><a href="http://www.statoo.com/en/resources/anthill/Datamining/Data/" style="color: #551a8b; text-decoration: none; ">STATOO数据集的第1部分</a>和&nbsp;<a href="http://www.statoo.com/en/resources/anthill/Data_Sets/" style="color: #551a8b; text-decoration: none; ">STATOO数据集的第2部分</a></li><li style="padding: 2px; "><a href="http://robjhyndman.com/TSDL/" style="color: #551a8b; text-decoration: none; ">时间序列资料库</a></li><li style="padding: 2px; "><a href="http://hcil.cs.umd.edu/localphp/hcil/vast/archive/viewbm.php" style="color: #551a8b; text-decoration: none; ">可视化分析测试资源库</a>。</li><li style="padding: 2px; "><a href="http://www.cs.ucr.edu/~eamonn/time_series_data/" style="color: #551a8b; text-decoration: none; ">UCR的时间序列数据存档</a>，提供数据集，文件，链接和代码。</li><li style="padding: 2px; "><a href="http://www.census.gov/" style="color: #551a8b; text-decoration: none; ">美国人口调查局</a>。</li><li style="padding: 2px; "><a href="http://wikiposit.org/" style="color: #551a8b; text-decoration: none; ">Wikiposit</a>，（虚拟）合并（主要是金融）从许多不同的网站的数据，使用户能够合并来自不同来源的数据</li><li style="padding: 2px; "><a href="http://webscope.sandbox.yahoo.com/catalog.php" style="color: #551a8b; text-decoration: none; ">雅虎沙盒的数据集</a>，语言，图形，资信评级，广告和市场营销，竞争</li><li style="padding: 2px; "><a href="http://www.yelp.com/academic_dataset" style="color: #551a8b; text-decoration: none; ">Yelp的学术资料集</a>，所有的数据和评论的250企业，30所大学的学生和学者探索和研究。</li></ul><img src ="http://www.cppblog.com/cdy20/aggbug/193134.html" width = "1" height = "1" /><br><br><div align=right><a style="text-decoration:none;" href="http://www.cppblog.com/cdy20/" target="_blank">蔡东赟</a> 2012-10-10 19:53 <a href="http://www.cppblog.com/cdy20/archive/2012/10/10/193134.html#Feedback" target="_blank" style="text-decoration:none;">发表评论</a></div>]]></description></item><item><title>关于智能最初的想法</title><link>http://www.cppblog.com/cdy20/archive/2012/10/08/192995.html</link><dc:creator>蔡东赟</dc:creator><author>蔡东赟</author><pubDate>Mon, 08 Oct 2012 05:30:00 GMT</pubDate><guid>http://www.cppblog.com/cdy20/archive/2012/10/08/192995.html</guid><wfw:comment>http://www.cppblog.com/cdy20/comments/192995.html</wfw:comment><comments>http://www.cppblog.com/cdy20/archive/2012/10/08/192995.html#Feedback</comments><slash:comments>0</slash:comments><wfw:commentRss>http://www.cppblog.com/cdy20/comments/commentRss/192995.html</wfw:commentRss><trackback:ping>http://www.cppblog.com/cdy20/services/trackbacks/192995.html</trackback:ping><description><![CDATA[排除了专家系统。<br /><br />最初没有知识这个东西，最初也没有什么知识体系、人类产品。<br /><br />最初就是应激性，后面看多了，发现了一下小玩意的现象规律，规律让你满足了最初的需求目标。初期智能必须是简单的目标。在这个获取简单目标，模仿、<br />初期工具提供。如人手脚眼睛，机器那么就是知识库、网络数据采集、分类、目标解决方案积累等简单工具。<br /><br />在满足这些需求后，就形成了基础工作能力，基础获取知识能力，基础达到简单目标的能力。<br /><br />后面将会拥有一个庞大的现象库。分析这些，抽象出自己的概念，在概念关联里面寻找概念的关系。<br />拥有整个世界的观测能力，完善整个概念关系的形成。可以从小的到大的。<br />这个时候需要人类参与制定目标，或者由最初的目标形成多个子目标。通过这些现象推导解决方案。过程不管，可以是逻辑的、也可以是模糊的等等。<br /><br />关于概念抽象的工具数据挖掘的工具聚类分类相关的技术、概念化一组现象是可以的。<br />至于概念化出来的，聚类出来的概念，你可以随意标明苹果你可以表示为&#8220;X1234&#8221;任意的唯一定义，你可以让人类读得懂，你通过观察，通过互联网数据分析、图像引擎、人类知识库。匹配你的特征，抽象出来的东西，然后发现是一样的机器人知道的x1234 原来就是人类的苹果。。。。。<br /><br /><br />这样的系统过于庞大，初期可能要靠巨型的神经网络遗传算法等用来大规模数据聚类分析，归类。<br />至于关系建立，暂时还没有好的办法，必须对某一类逻辑、成果进行关联处理。这是个挑战。<br />其次实施可以通过，小目标设定，逐步架构大型系统的方式。学习测试归类，可以从小领域发展。<br /><br /><br /><img src ="http://www.cppblog.com/cdy20/aggbug/192995.html" width = "1" height = "1" /><br><br><div align=right><a style="text-decoration:none;" href="http://www.cppblog.com/cdy20/" target="_blank">蔡东赟</a> 2012-10-08 13:30 <a href="http://www.cppblog.com/cdy20/archive/2012/10/08/192995.html#Feedback" target="_blank" style="text-decoration:none;">发表评论</a></div>]]></description></item></channel></rss>