How Should I Represent My Input with Multiple fields?

[rhyolight]

@sheiser1

My question is, what’s the advantage(s) ofdoing this rather than increasing the size of the encoder vector, to say2000 instead of 400 (by giving each field a sub-encoding of n=200 andw=15), and using local inhibition to handle the larger encoding vector?

The bigger the input space of the spatial pooler, the longer it takes to process the input. I’m trying to keep the input smaller. We know from experience that just 4 scalar input fields (1600 bits) can slow the SP down quite a bit. Local inhibition slows things down a lot more, and doesn’t make sense to use unless the input data is topological.

@sheiser1

I remember you mentioning in anotherpost that imposing topology w/local inhibition had a negative effect onresults and I’m curious what may be lost doing it this way?

The SP will be looking for topological associations in the input space where there are none.

@sheiser1

It seems tome that single regions higher up the cortical hierarchy may be looking atmultiple sub-regions, meaning that their SP’s may be dealing with largerencoding vectors.

Remember we are not dealing with hierarchy yet.