On Mon, Feb 18, 2019 at 4:01 PM Ben Goertzel <b...@goertzel.org> wrote:
> *** > ... > And likely the way to do this is to set the network oscillating, and > vary inhibition to get the resolution of "invariants" you want. > *** > > But we are not doing that. Interesting... Cool. Maybe there could be a match. I want hardware to try this. I've been playing with the open SpiNNaker interface with the European Human Brain Project. But it's batch jobs, and lots of fiddling with Python interfaces for neurosimulators. I'd love to collaborate on it. I've always thought the set based representations Open Cog was using could be fitted to what I was doing. But I didn't realize you were doing anything like clustering raw sequence networks on the fly for language. As I recall when I talked to Ruiting in 2013 she was using rules?? I only started looking in detail at networks after 2013. I was trying to fit it to Jeff Hawkins' sequence networks. I finally figured out the whole thing, even the "cross-product" in my original formulation, would reduce to something as simple as diamond shaped "cliques" in the network. But I didn't know how to isolate them. Then I came across this paper: A Network of Integrate and Fire Neurons for Community Detection in Complex Networks Marcos G. Quiles, Liang Zhao, Fabricio A. Breve, Roseli A. F. Romero http://www.sbmac.org.br/dincon/trabalhos/PDF/invited/69194.pdf I immediately googled to see if there was any evidence this mapped to perception experimentally. And found "binding by synchrony" has been observed since the '80s, but no-one knew why! Nice for me, because I'm working from the other direction. I have the network, and the network predicts binding by synchrony. If you want to compare notes and see if there are any cross insights which could inform, we can talk over email if you like. -Rob ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T581199cf280badd7-M11ab0b41875929fa3b7d0bee Delivery options: https://agi.topicbox.com/groups/agi/subscription