9 replies to this topic

[YOLF]

So how can I get started? Is there something that's easy to start and can take me a long way with AI? I'd eventually like to use AI to sort and arrange information for my reading/educational pleasure.

Something that will run on multiple platforms is good too, I want it to be available from any machine. Having it run in a web browser would be cool too.

Edited by cryonicsculture, 03 April 2013 - 05:46 AM.

[Avatar of Horus]

I am new too to the AI field, and want to learn it for similar reasons, to help with my studies and research, etc., so I can't mention something specific for starters, but I'd suggest Wikipedia, it's possibly good for a starting point, like this:
wikipedia/Artificial_intelligence
and some related:
wikipedia/Intelligent_agent
wikipedia/Knowledge_representation
and especially this one:
wikipedia/Ontology_(computer_science)

The point is that one needs to the basic concepts and then an interrelated network based on them to represent the knowledge.

Edited by Avatar of Horus, 07 April 2013 - 03:04 PM.

[YOLF]

Thanks!

[Deeviant]

AI is such a broad field, that I think you may have to pick a more specific topic within to gain traction. Also, I'm not sure how much knowledge you have in the computer science field in general, but basic foundational computer science knowledge, such as data structures, design patterns and common algorithms is an obvious place to start.

Your project you mentioned, for instance, a system that would gather and sort information that you would find interesting could be approached many different ways. For instance, you could attempt to create some sort of AI system that functionally aware of if interests and possess an effective hueristic that is able to judge new material, classify it and organize it, but frankly, the scope of that project is very large and perhaps insurmountable.

A more pragmatic programmer might appoach it differently. For instance, they might utilize existing webscrapping frameworks, and use a statistical approach that uses keywords, associations and popularity to generally get at some useful and topical information; which is a very simply description of what Google does with their search. This is not an "AI".

I don't think there is an "easy" why to get into AI. It has baffled the best computer scientists of the last 60 years, after all. But I think researching the many different branches of the field, and choosing one to focus on, is probably the best first step.

[Avatar of Horus]

I found a related relevant topic, here:
Introduction to AI
http://www.longecity...oduction-to-ai/

...
I don't think there is an "easy" why to get into AI. It has baffled the best computer scientists of the last 60 years, after all. ...

You've put it right I think. Cf.:

AI is one of the newest fields in science and engineering. Work started in earnest soon after World War II, and the name itself was coined in 1956. ...
A student in physics might reasonably feel that all the good ideas have already been taken by Galileo, Newton, Einstein, and the rest. AI, on the other hand, still has openings for several full-time Einsteins and Edisons.

Source: Artificial Intelligence - A Modern Approach, Third Edition, pg. 1

Edited by Avatar of Horus, 17 February 2014 - 04:12 AM.

[Sergej Shegurin]

See courses "neuralnets" by Geoffrey Hinton, "ml" by Andrew Ng, "nlangp" at Coursera.
Also IPAM 2012 graduate school on deep learning is very useful. The very brightest stars of AI field teach there like Geoffrey Hinton, Yann LeCun, Ranzato, Bengio and others.

Edited by Sergej Shegurin, 24 February 2014 - 05:12 PM.

[Avatar of Horus]

Here is a web page, from the author of the aforementioned book, called:
AI on the Web
http://aima.cs.berkeley.edu/ai.html
which lists links to hundreds of pages around the web with information on Artificial Intelligence.

For example:

WHAT IS ARTIFICIAL INTELLIGENCE?
Abstract:
This article for the layman answers basic questions about artificial intelligence. ...

http://www-formal.st...i/whatisai.html

[Olon]

Sorry for threadjacking, but I find the question doesn't justify an extra thread:
Where can I find large numbers of very low resolution (16*16 or 32*32 pixels, black and white) pictures, I fear when I create own ones out of photographs they will no longer be recognizable. Reducing the resolution of pictograms should be easy, but I also wanto to have more natural-looking pictures.

Edited by Olon, 19 March 2014 - 05:23 PM.

[lemonhead]

I read Philip Johnson-Laird's The Computer and the Mind a while back and liked it. I was hesitant to suggest it, but given your recent question you might be interested. It has a good section on vision.

Edited by lemonhead, 20 March 2014 - 12:38 AM.

[LexLux]

10 online courses:

http://efytimes.com/...asp?edid=127676