@liyzhen2 Hopefully that answers your question! Kull & Flach (2015): https://t.co/w5oxbSiUGj Lahlou et al. (2021): https://t.co/clXQnZ9QpD
1,923 followers
1,695 followers
RT @octonion: It appears log-loss can be decomposed, just like Brier score. Novel Decompositions of Proper Scoring Rules for Classificatio…
21,763 followers
It appears log-loss can be decomposed, just like Brier score. Novel Decompositions of Proper Scoring Rules for Classification: Score Adjustment as Precursor to Calibration https://t.co/AX4fvp64FA
14,941 followers
@Abel_TorresM @neurowitz Finally, there are papers with multiple local loss functions that combine to provide a global guarantee. This was the goal of difference target propagation, for example: https://t.co/ePglgz901O