<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: Captcha cracking in JavaScript with Canvas and neural nets</title>
	<atom:link href="http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/feed" rel="self" type="application/rss+xml" />
	<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets</link>
	<description>Cleaning up the web with Ajax</description>
	<lastBuildDate>Thu, 17 May 2012 07:43:39 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.3.2</generator>
	<item>
		<title>By: sourceRoot</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-284351</link>
		<dc:creator>sourceRoot</dc:creator>
		<pubDate>Thu, 24 Jun 2010 23:23:35 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-284351</guid>
		<description>The &quot;spam question&quot; is far worse at stopping spam than captcha :) 

Most of the effective spammers use low wage labor to post somehting kinda on topic like nice post I agree with so and so... link

Captchas have been dead for a long long time</description>
		<content:encoded><![CDATA[<p>The &#8220;spam question&#8221; is far worse at stopping spam than captcha :) </p>
<p>Most of the effective spammers use low wage labor to post somehting kinda on topic like nice post I agree with so and so&#8230; link</p>
<p>Captchas have been dead for a long long time</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Bigbx</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-275080</link>
		<dc:creator>Bigbx</dc:creator>
		<pubDate>Fri, 14 Aug 2009 19:25:12 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-275080</guid>
		<description>Why complicate when there ready firefox plugin for this? Use google and spend time on smarter things.</description>
		<content:encoded><![CDATA[<p>Why complicate when there ready firefox plugin for this? Use google and spend time on smarter things.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: ilazarte</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270746</link>
		<dc:creator>ilazarte</dc:creator>
		<pubDate>Mon, 26 Jan 2009 17:10:20 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270746</guid>
		<description>I guess this marks the end of captchas!  

Welp, let me just finish my post here, by answering this Spam Question located at the bottom of this form...</description>
		<content:encoded><![CDATA[<p>I guess this marks the end of captchas!  </p>
<p>Welp, let me just finish my post here, by answering this Spam Question located at the bottom of this form&#8230;</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: ThomasHansen</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270739</link>
		<dc:creator>ThomasHansen</dc:creator>
		<pubDate>Mon, 26 Jan 2009 14:19:12 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270739</guid>
		<description>@Snyke and ffreak
You divide the image into a n matrices where n is the number of unique elements you find in it (characters) then you treat every matrix as a vector and check the &quot;direction&quot; according to your precompiled &quot;solution&quot;...
Normally *WAY* faster then NN and doesn&#039;t lock up or any of the other &quot;great problems&quot; that NN comes with...
.
It&#039;s quite common according to my knowledge for OCR and similar things...
.
not sure how good it would work for captchas though since they tend to &quot;mess&quot; with the surface making it more difficult to recognize the pattern...
.
Don&#039;t get me wrong, I think NNs in JavaScript sounds hilariously cool, though I think the Linear Algebra solution would perform *seriously* faster (if possible)
.
Beside it was a *question* not an attack ... ;)</description>
		<content:encoded><![CDATA[<p>@Snyke and ffreak<br />
You divide the image into a n matrices where n is the number of unique elements you find in it (characters) then you treat every matrix as a vector and check the &#8220;direction&#8221; according to your precompiled &#8220;solution&#8221;&#8230;<br />
Normally *WAY* faster then NN and doesn&#8217;t lock up or any of the other &#8220;great problems&#8221; that NN comes with&#8230;<br />
.<br />
It&#8217;s quite common according to my knowledge for OCR and similar things&#8230;<br />
.<br />
not sure how good it would work for captchas though since they tend to &#8220;mess&#8221; with the surface making it more difficult to recognize the pattern&#8230;<br />
.<br />
Don&#8217;t get me wrong, I think NNs in JavaScript sounds hilariously cool, though I think the Linear Algebra solution would perform *seriously* faster (if possible)<br />
.<br />
Beside it was a *question* not an attack &#8230; ;)</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Snyke</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270717</link>
		<dc:creator>Snyke</dc:creator>
		<pubDate>Sun, 25 Jan 2009 20:04:23 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270717</guid>
		<description>Agreed with ffreak, there is currently no better option to NN for pattern recognition.
Still it would be interesting to see how this works with recaptcha which is a project for using distributed users entering captchas to digitalize old scriptures.</description>
		<content:encoded><![CDATA[<p>Agreed with ffreak, there is currently no better option to NN for pattern recognition.<br />
Still it would be interesting to see how this works with recaptcha which is a project for using distributed users entering captchas to digitalize old scriptures.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: ffreak</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270716</link>
		<dc:creator>ffreak</dc:creator>
		<pubDate>Sun, 25 Jan 2009 15:47:29 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270716</guid>
		<description>@ThomasHansen
Because patterns recognition is what neural networks do great and they are easy to implement. I have no idea how anyone could use &quot;linear algebra&quot; instead - I suppose it would be insanely complex.</description>
		<content:encoded><![CDATA[<p>@ThomasHansen<br />
Because patterns recognition is what neural networks do great and they are easy to implement. I have no idea how anyone could use &#8220;linear algebra&#8221; instead &#8211; I suppose it would be insanely complex.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: ThomasHansen</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270715</link>
		<dc:creator>ThomasHansen</dc:creator>
		<pubDate>Sun, 25 Jan 2009 11:59:18 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270715</guid>
		<description>Impressive!
Though why not use Linear Algebra instead of Neural Networks...?
I assume NN have way higher processing demands and still less right hits...?
Comments...?</description>
		<content:encoded><![CDATA[<p>Impressive!<br />
Though why not use Linear Algebra instead of Neural Networks&#8230;?<br />
I assume NN have way higher processing demands and still less right hits&#8230;?<br />
Comments&#8230;?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Jordan1</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270714</link>
		<dc:creator>Jordan1</dc:creator>
		<pubDate>Sun, 25 Jan 2009 01:52:55 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270714</guid>
		<description>Wow its even better than the best PHP captchas! They should put it in one of those annual best captcha competitions. Its inevitable that neural networks would one day surpass human perception.</description>
		<content:encoded><![CDATA[<p>Wow its even better than the best PHP captchas! They should put it in one of those annual best captcha competitions. Its inevitable that neural networks would one day surpass human perception.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: nunziofiore</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270711</link>
		<dc:creator>nunziofiore</dc:creator>
		<pubDate>Sat, 24 Jan 2009 17:14:33 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270711</guid>
		<description>lucky ajaxian login goes again :)

this is a bit complexed but finally i understood it... great...
anybody know something or can suggest some link about MooTools+CANVAS+[IPHONE&#124;ANDROID] ?
Does the browser mounted on that device support it?</description>
		<content:encoded><![CDATA[<p>lucky ajaxian login goes again :)</p>
<p>this is a bit complexed but finally i understood it&#8230; great&#8230;<br />
anybody know something or can suggest some link about MooTools+CANVAS+[IPHONE|ANDROID] ?<br />
Does the browser mounted on that device support it?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: oopstudios</title>
		<link>http://ajaxian.com/archives/captcha-cracking-in-javascript-with-canvas-and-neural-nets/comment-page-1#comment-270708</link>
		<dc:creator>oopstudios</dc:creator>
		<pubDate>Sat, 24 Jan 2009 11:44:18 +0000</pubDate>
		<guid isPermaLink="false">http://ajaxian.com/?p=5786#comment-270708</guid>
		<description>Excellent read! But it begs the question, are there systems that could crack a form that is *activated* with javascript (thus the source doesn&#039;t represent the true form)? I&#039;m not up-to-date with the modern spiders...</description>
		<content:encoded><![CDATA[<p>Excellent read! But it begs the question, are there systems that could crack a form that is *activated* with javascript (thus the source doesn&#8217;t represent the true form)? I&#8217;m not up-to-date with the modern spiders&#8230;</p>
]]></content:encoded>
	</item>
</channel>
</rss>

