A continuous exact reformulation lets precision, recall, and F-beta metrics be optimized with gradient methods, avoiding smooth surrogate losses.
Belkin, Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolation, Acta Numerica, 30 (2021), pp
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Exact Reformulation and Optimization for Direct Metric Optimization in Binary Imbalanced Classification
A continuous exact reformulation lets precision, recall, and F-beta metrics be optimized with gradient methods, avoiding smooth surrogate losses.