Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
For the out-of-band Slashdot experience (mostly headlines), follow us on Twitter, or Facebook. ×

+ - SimpleAI, an implementation of artificial intelligence algorithms->

machinalis writes: "SimpleAI is an easy to use lib implementing in python many of the artificial intelligence algorithms described on the book "Artificial Intelligence, a Modern Approach", from Stuart Russel and Peter Norvig.
This implementation takes some of the ideas from the Norvig's implementation (the aima-python lib), but it's made with a more "pythonic" approach, and more emphasis on creating a stable, modern, and maintainable version. We are testing the majority of the lib, it's available via pip install, has a standard repository and lib architecture, well documented, respects the python pep8 guidelines, provides only working code (no placeholders for future things), etc. Even the internal code is written with readability in mind, not only the external API.
This new release adds a few statistical classification methods to SimpleAI with
the intention of start replicating the machine learning aspects of aima-python, also includes lots of tests for the classifiers, documentation, and a few sample uses of the classifiers.
twitter: @machinalis""

Link to Original Source

Comment: In Argentina... (Score 1) 168 168

There's a project called InfoLeg Trying to do exactly this... I'm not sure if they've been able to keep up, but they have a lot of content browsable in a way that resembles revision control a lot.

As an example you have here our copyright law passed in 1933. The three links there can show you the original text, the most recent text applying all updates (i.e. "HEAD" :) ), and a link to a list of laws/decrees that updated modified it later.

That's more or less what you were looking for? I know people who were close to the project in the past, I can get you a couple of names if you're really interested...

Comment: Pyweek (Score 1) 324 324

A lot of developers I've known had made their first game during Pyweek contests. Pyweek is a free and open game creation contest, using python. It has a very friendly and open community, so even if it is a contest there is a lot of people around wanting to help newbies and provide advice.

But the best thing of participating in the contest is that the rules help you to FINISH a game. Starting work on a game is easy, but it's too tempting to fall into scope creep and start adding characters, places, game mechanisms, enemy behaviour ad infinitum, and you are always starting new stuff but never getting to have something finished.

Try it. Most of the people I known to go into the contest have had a lot of fun.

Comment: Nice extrapolation (Score 1) 867 867

They imagined 2.5 megawatt turbines crisscrossing the terrestrial globe, excluding 'areas classified as forested, areas occupied by permanent snow or ice, areas covered by water, and areas identified as either developed or urban,'

I hope the power is enough to make all the food replicators work. Otherwise I don't know what we will eat when we cover every arable field (read: the places where we grow most food now, which are not forested, with ice, water, nor urban) with wind turbines.

Comment: Re:Isn't this a dupe? (Score 1) 613 613

You want a solution? How about this: Windows should only hide file extensions for files that don't use custom icons. IOW, a .doc would show up as a Word document (by icon), so it doens't need the .doc. But if you change the icon of your .exe file to be the word doc icon, then the .exe still shows up.

I have an easier, better one: executability is not part of the file type. Use extensions to indicate file type (and hide it or show it at leisure), but put executability as an extra attribute of the file in the filesystem.

If it sounds familiar, it's because that's what UNIX has been doing since the '70s

Comment: Re:How can this be? sufixication (Score 2, Insightful) 613 613

To solve this security issue (malicious execution), you don't need separate forks, complex metadata, or anything like that. You just need a single bit of metadata, telling you if the file is executable or not. In other words (Henry Spencer's, not mine): Those who do not understand UNIX, are comdemned to reinvent it, poorly. With that metadata, the worst that somebody can do is name a file foo.txt.jpg to trick you into opening an image making you think it's a text file, but nothing that makes a security risk.

Comment: Perhaps using HTTP referrers? (Score 1) 297 297

It is hard to think a good solution without knowing what you mean by "locals" (is there a simple offline way to get them a passphrase?), what kind of users you have (will they be thrown off from the site if they have to login?), etc.

Anyway, something that might work and can be used together with other filtering methods mention by other people here is filtering by HTTP referrer field.

This is assumming that your users have alternate ("local") ways of getting the URL of your site, and you don't mind being a bit detached from the rest of the web (apparently, you don't). So what you do, is filter every HTTP request where the referrer field is not empty or inside your site. This will essentially stop people from following links from other sites (google, blogs, whatever) into your site. You will be able to enter your site only by entering your URL (like knowing the BBS phone number), and then following internal links. Of course, this is not "secure" (for someone who knows http, it is not very hard to enter directly once they realize how the filter works), but I think it will stop a lot of "casual bypassers".

Btw, if you avoid being in google (robots.txt covering your whole site) you will avoid a lot of visitors already.

"Silent gratitude isn't very much use to anyone." -- G. B. Stearn