Deep learning can’t be trusted, brain modelling pioneer says

Re:Deep learning can’t be trusted, brain modelling pioneer says by amelius on Thursday January 06, 2022 @05:44PM (#29831581)

He formulated his theory in the 70s, and it didn't catch on. How likely is it that he is right about this?

[ Parent | on HN ]

Re:Deep learning can’t be trusted, brain modelling pioneer says by UncleOxidant on Thursday January 06, 2022 @05:59PM (#29831780)

The roots of deep learning are from around the same era, if not a bit earlier - it's just that we didn't have the computing power to do much with it so DL kind of languished until ~15 years ago with the advent of capable GPUs. Is ART important? Can it do what DL does now? Hard to say. A lot of money and effort has been poured into DL to get it this far. If ART had gotten the attention/investment instead would it be where DL is now?

[ Parent | on HN ]

Re:Deep learning can’t be trusted, brain modelling pioneer says by Den-vr on Thursday January 06, 2022 @05:48PM (#29831635)

Many excellent things were formulated in the 70s and didn't catch on because supporting technologies weren't practical yet. It was an excellent time to be a fresh PhD

[ Parent | on HN ]

Re:Deep learning can’t be trusted, brain modelling pioneer says by jjtheblunt on Thursday January 06, 2022 @06:09PM (#29831920)

it was also an excellent time to be a really little kid, with so much to marvel at (which is really to your point!)

[ Parent | on HN ]