News for hackers. Stuff that matters.
Re:At the Far Ends of a New Universal Law by gone35 on Wednesday October 15, 2014 @06:54PM (#8462310)
Thank you for sharing this! Not my field at all [1], but reading through your result, I cannot help but think of L1-regularization in ensemble optimization/learning algorithms; see for instance Section 6.3 in [2]. Perhaps it's too leaky of a heuristic, but it definitely struck a chord.

[1] http://xkcd.com/793/

[2] http://face-rec.org/algorithms/Boosting-Ensemble/8574x0tm63n...

Re:At the Far Ends of a New Universal Law by rdlecler1 on Wednesday October 15, 2014 @09:29PM (#8462818)
Thanks for sharing, I need to dive into that deeper when I have some relaxation time!

It's funny that you mention that because the mathematics used to describe the dynamics of artificial neural networks are the same mathematics used to describe (artificial) gene networks. In effect, a gene network is the 'brain' of the cell. In fact, it was such an effective system that life re-evolved this computational architecture with a different substrate: neurons. Functionalism at its finest! Insofar as cells communicate with one another and are composed of the exact same gene networks (albeit in different states), you basically have one giant meta neural network comprised of identical 'tiles' neural networks each 'tile' connected to their nearest neighbors.