On Sat, Nov 9, 2024 at 1:59 AM Danko Nikolic <danko.niko...@gmail.com> wrote:
> ... gating networks have to learn cumulatively, starting from simple > things and gradually expanding their knowledge, one step at a time (again, > this is how biological intelligence learns, so gating stays consistent with > biological intelligence in that respect too). ... adding more neurons and > gates when the current system hits a wall, and so on > Mentioning "neurons" sounds like the opposite of brain development even for newborns. Although it is true that newborns have a few thousand connections per neuron and that dramatically increases so that by age 2-3 they have ~15k connections per neuron, the rest of development is net-pruning till adults have half as many connections. If I were serious about modeling brain development, I'd treat the increase of connections from birth to infancy as mainly attempting to get the brain out of the birth canal before going for lottery ticket learning: First dramatically and relatively randomly increase the number of connections so that real learning by pruning could commence with winning lottery tickets. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T232f2f385a5389ed-M3e258d45da9850be074b00b9 Delivery options: https://agi.topicbox.com/groups/agi/subscription