4 ms·
The purpose of post-prediction explanations would be to increase confidence of a practitioner to use said inference. It’s a disconnect between finding a real l
by Narhem 2y ago
The purpose of post-prediction explanations would be to increase confidence of a practitioner to use said inference.
It’s a disconnect between finding a real life “AI” and trying to find something which works and you can have a form of trust with.
- solidninja 2y agoIs there a study of "smooth"/"stable" "AI" algorithms - i.e. if you feed them input that is "close" then then the output is also "close"? (smooth as in smoothly differentiable/stable as in stable sorted)