I’m not easily impressed but *this* impresses me.
I expected it to notice “interesting things happening” in natural cavities… but no, they took it much further and mathematically modeled and have a proper glossary to describe structure. [I’ll give example in the comments]
I know this means nothing to most people on my friends list but *this* here is one of the missing links in properly modeling real neural networks because they *include* the structures themselves in how they’re found and behave.
It’s not just weights but how the weights are located topographically that makes a difference.
Think of putting a set of earbuds in your pocket – the ones with the wires.
Then you take them out and they’re knotted together.
Sort of like that. Not really like that but sort of like that.
Why this impresses me: this creates a language to describe mathematically different structures neurons may fall into (or grow out of) – AND it’s using mathematics while *also* being linked to real biology.