Artificial Intelligence Depot
Visiting guest. Why not sign in?
News, knowledge and discussion for the AI enthusiast.
FEATURES COMMUNITY KNOWLEDGE SEARCH  
Developers, get involved!
The AI Foundry promotes the development of Open Source artificial intelligence projects. Why not join it now? All projects welcome extra help!
Visit the AI Foundry

Reply to Message

Not registered yet?

The AI Depot has a focused community of friendly users. Rather than let anyone abuse the site at the brink of promiscuity, we prefer to let only those with an active interest participate... this simply requires registering.

Why not sign up!

Joining the site's community is completely free. You can then post messages freely, and customise your personal profile at will. Specific privileges will also be granted to you, like being able to access printer-friendly articles without restrictions. So, why not register?

Username:
Password:
Subject:
Email me when someone replies.
Body:

Parent Message

Category ART

Feel free to have a look through the source code provided on the applet pages. I'm afraid that I never really got around to commenting them better but the algorithms are quite simple. David Weenik's papers on his modified ART algorithms easy to follow if you'd rather code from the original source. In addition, you might consider using Growing Neural Gas algorithms that are also mentioned on the applet pages.

-RoB-

47 posts.
Sunday 02 June, 10:05
Reply
Classifying visual inputs

I had a look through your code and I've modified Rodney's STVM accordingly. I do remember seeing a video a long long time ago where (I think) Grossberg described his ART algorithm. It was a long boring highly mathematical lecture and I never payed much attention at the time. But reading your paper it seemed like the classification version of ART would be a neat solution to classifying the robots visual input. The expanding and contracting tollerances (receptive fields) mean that the robot can produce fine classifications for objects which are more interesting or more frequently observed, and fairly broad classifications over areas of the feature space which aren't very interesting (such as Grossberg lectures).

I don't know exactly how you did it but I just wrote an algorithm such that each "neuron" tries maintain a constant number of neighbours within its receptive field. If there are not enough neighbours the field expands. If there are too many neighbours the field contracts.

- Bob

136 posts.
Sunday 02 June, 17:41
Reply
Good idea

Bob, Trying to maintain a constant number of neighbours per neuron sounds very reasonable. One of my research aims was to show that a model of curiosity could be built with standard components so I just implemented the algorithm as published -- reducing the receptive field by a fixed percentage (eg. 10%) as needed. I'd be interested to know how you get on with your variation. -RoB-

47 posts.
Monday 03 June, 16:18
Reply

Back to the Artificial Intelligence Depot.