I was learning self organizing feature maps the other day. I want to intuitively understand it because I'm not that good at math. But I still am not very clear about it. I can easily understand choosing a winner neuron and adjusting it's weights according to the sample. But what I cannot understand is the idea of the neighbourhood.
Why not update the winner, the neuron with the closest value to the sample, alone? What is the importance of updating close by neurons with it?
And why should you reduce the neighbourhood as you go on? Why not keep the neighbourhood just fixed at a value?