Re:The revolution of machine learning has been exaggerated
by pradn on Friday November 22, 2019 @06:03PM
(#21611049)
It's also a problem that students are focusing so much on neural-network-based approaches, neglecting other techniques. There's what, like, 10 places in the world where you'll have enough compute to be able to train cutting-edge neural networks? Real world problems are just as well solved with other techniques like XG-Boost, which wins pretty often in Kaggle, for example.
[ Parent
| on HN ]
Re:The revolution of machine learning has been exaggerated
by hnaccy on Friday November 22, 2019 @08:28PM
(#21611809)
>There's what, like, 10 places in the world where you'll have enough compute to be able to train cutting-edge neural networks
You can just use a smaller amount of compute with transfer learning style stuff.
I can probably fine-tune a transformer model with my pocket money and beat any NLP solution from two years ago.