Re:Ask HN: Why companies are not using deep learning yet?
by natalyarostova on Friday February 15, 2019 @05:33PM
(#19175349)
To elaborate on your point, it doesn't just require lots of data, it requires lots of data with a corresponding high effective sample size. If you want to forecast sales for next Christmas, I don't care if you have 2000 terabytes of granular orders and sales data, because the effective sample size for past observed Christmases is going to be like 3-4 (Christmases further than 4 years back may no longer be representative).
In these cases nice structural time-series models, which are in spirit not so different from what existed 20 years ago, will beat deep learning.
Re:Ask HN: Why companies are not using deep learning yet?
by avin_regmi on Friday February 15, 2019 @07:52PM
(#19176227)
What frameworks are you using currently in production? Do you think classical machine learning algorithms are being used more than deep learning?
Deep learning requires lots of data, and at best, it's about as effective as a dumb foreign worker. It's not going to replace any jobs soon and it's a very long game.
Also if you train it on garbage data, you get garbage results. Not everyone has access to clean data. When many people say data is the new oil, really they're just putting a mountain into a blender and expecting deep learning to find the oil.