Tutorial - Self-Organising Maps For Colour Recognition

Max Dennis Luesebrink's contest entry has just been published.

Looking into unsupervised learning techniques, this article offers a different approach to recognising colours. The system can train itself to match similar colours together. When asked about unknown colours, it learns them for future queries.

http://ai-depot.com/Tutorial/SomColour.html

Hi,

Can your system be used to recognise hairs coming from same persons?
I have to work in that field and I conclude that it's a real challenge.
Regards,
Alain

Hello Hubrecht

Of course you have to know what your input of the hairs is. Or do you put that in an apparatus or something like that. But then that apparatus produces output, which is your input for recognizing. Or don't I understand you?

A neural net that works with margins can recognize all you want. In a goof net it's not necessary to learn first a reference. You get the different types of hairs grouped, if you want that.

But if I was you I would extend the question somewhat so more people can react. And possibly you will get then a more precise answer as well.

It's indeed a nice application of self-organizing.

Ed van der Meulen