Evolution Is the New Deep Learning

Post by cshenton » Wed, Mar 28, 2018, 1:27 AM UTC

make3 wrote: post #224

cshenton wrote: post #130

Earlier quoted context omitted.

Actually, evolutionary strategies make use of either empirical or naturalised gradients on the search space.

I guess sexual reproduction is kind of like sharing gradients. Is that what you mean?

Not quite, genetic algorithms represent the search space usually as bit sequences, so you can just mix subsequences from two parents, that's reproduction.

In contrast, evolution strategies have a search distribution, and compute a gradient update to that distribution from an entire generation. So it's really more 'search gradients' than evolution. The only thing 'evolutionary' about it is that the minibatches are called generations.

Post by jcora » Fri, Mar 30, 2018, 7:37 AM UTC

dorgo wrote: post #192

jcora wrote: post #154

Earlier quoted context omitted.

You should study ML a bit more, to see just how wrong you are about it

I said two things. Which am I wrong about?

1. We haven't 'figured out' intelligence, far from it, we don't even know what we don't know.

Your comment sounds like Lord Kelvin proclaiming that physics is "over" a couple years before people figured out there were huge holes in the theory which eventually led to quantum mechanics. Our understanding of intelligence is probably less complete than our understanding of physics _back then_.

2. ML is really underwhelming if you measure it against actually intelligent behavior. Figuring out cool regression mechanisms is neat and all, but that's what it is, and it has nothing to do with intelligence in the general sense, much like expert systems had nothing to do with actual domain knowledge, they were just one of the most primitive models, low-hanging fruit that we could exploit.

Re: Evolution Is the New Deep Learning

Re: Evolution Is the New Deep Learning