This approach to ML seems very similar to HTM

Science Tangential Theories
#1

They are modeling the neuron as predicting its own input in place of traditional back prop. I really enjoyed the talk.

3 Likes
#2

I talked to Blake a bit about this in this podcast.