Topographic Infomax in a Neural Multigrid

Ed_Pell · July 23, 2018, 1:55am

Is this using grid cell arrays as a spacial basis, a multigrid, for infomax?

Sean_O_Connor · July 23, 2018, 12:58pm

It seems like a nice idea trapped behind a pay-wall. It’s kind of like what Nvidia did with GAN neural networks, gradually adding in higher frequency components during training.
I guess Numenta could FFT sequence data to create a low pass filter and then gradually add in higher frequency components during training.

Ed_Pell · July 23, 2018, 5:05pm

I did not realize this was behind a paywall sorry about that.

Bitking · July 23, 2018, 5:34pm

You can request a freebie copy here:
https://www.researchgate.net/publication/220870168_Topographic_Infomax_in_a_Neural_Multigrid

I did and will follow-up if I get one.

Topic		Replies	Views
Spatial Pooling, SDRs, and Grid Cell Dimensionality Phase Transitions Numenta Theory	9	1019	January 8, 2020
HTM Mini-Columns into Hexagonal Grids! Tangential Theories grid-cells , hexagonal , binding	159	15413	March 11, 2020
Output cell activation information of two layers as sparse distributed input to higher layer Numenta Theory hierarchy	0	627	November 11, 2016
HTMs as Serial Associative Memories Getting Started question	4	525	May 5, 2020
HTM Learning algorithm Related Papers	2	973	November 3, 2017

Topographic Infomax in a Neural Multigrid

Related Topics