Re:Some thoughts on machine learning with small data
by figurative on Friday June 03, 2022 @06:55AM
(#31606686)
Just out of curiosity where do you see the demand? I've also heard from others that there's been a shift lately towards simpler models. Combined with domain knowledge linear/logistic regression can be really impressive!
[ Parent
| on HN ]
Re:Some thoughts on machine learning with small data
by rmnclmnt on Friday June 03, 2022 @08:08AM
(#31607181)
Small non-FAANG companies usually, where they do not have the internal skills to maintain and explain models. And from what I've heard, big corporations under regulated domains (banking, healthcare, etc).
Re:Some thoughts on machine learning with small data
by dbs on Friday June 03, 2022 @07:23AM
(#31606875)
> Combined with domain knowledge linear/logistic regression can be really impressive!
Can you elaborate?
Re:Some thoughts on machine learning with small data
by rmnclmnt on Friday June 03, 2022 @08:13AM
(#31607222)
I can agree this the comment. Linear models combined with advanced feature engineering gathered from domain knowledge can achieve great results in a white-box fashion!
A nice keynote by Vincent Warmerdam [1] talks about tips and tricking for advanced feature engineering combined with linear models.
Re:Some thoughts on machine learning with small data
by beckingz on Friday June 03, 2022 @11:23AM
(#31609565)
A significant portion of ML workloads involve predicting or classifying something. Linear/logistic regression of the right variables/features typically gets a significant portion of the data's ability to predict /classify correctly, while being significantly easier to build, train, deploy, and understand.
Heck, in a large number of domains, simple ratios -- debt to income ratio in finance for example -- will dominate the feature weight for many models and can be used on their own as a pretty good heuristic.