So after my usual bouts of indecision, here's my paper for tomorrow:
http://www.cs.utoronto.ca/~hinton/absps/ijcai05.pdf
Sorry to post it so late.
Apparently using neural networks to model handwriting recognition is not a very new thing, but:
1) it's new to me! i think this paper could make a nice link in our discussion between free energy minimisation and Michael's words on Bayesian theory.
2) this particular approach uses multiple neural networks to model the recognition of finer and finer details in handwriting. from what i understand, THIS is a bit new. also, there is a great Flash demo online. I just have to find it again and i'll post it up!
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment