Nested Learning: A new ML paradigm for continual learning

(research.google)

3 points | by gmays 9 hours ago ago

1 comments

  • Marshferm 9 hours ago

    Memory in animals is chained in still unknown ways along topological lines probably as affinities of commonality. It’s both hidden, discrete and continuous.

    Current ideas of computer “learning” are so impoverished it’s like an erector set being used for bridge construction. Start at analog units, forget this binary overbuild.