Re:A Short Chronology of Deep Learning for Tabular Data
by Tenoke on Monday September 05, 2022 @05:42AM
(#32722308)
This is such an odd dogmatic generalization. There's plenty of business cases where e.g. SHAP is enough for explainability and having a higher accuracy is important.
[ Parent
| on HN ]
Re:A Short Chronology of Deep Learning for Tabular Data
by dr_dshiv on Monday September 05, 2022 @07:15AM
(#32723041)
Yeah, but if you don’t know why you have high accuracy, it is hard to feel confident that your accuracy will remain high.
What is SHAP?
Re:A Short Chronology of Deep Learning for Tabular Data
by Tenoke on Monday September 05, 2022 @08:13AM
(#32723401)
SHAP is a recent de-facto standard way to get feature importance out of any model, whether you care about how much each column in your dataset contributes to a given decision or say which pixels in an image mattered the most.