A new loss term that optimizes the likelihood of any class being present improves multi-label classification with abundant negative data, with gains up to 6.01 points in F1.
Toward purifying defect feature for multilabel sewer defect classification.IEEE Transactions on Instrumentation and Measurement, 72: 1–11, 2023
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
unclear 1representative citing papers
citing papers explorer
-
Any-Class Presence Likelihood for Robust Multi-Label Classification with Abundant Negative Data
A new loss term that optimizes the likelihood of any class being present improves multi-label classification with abundant negative data, with gains up to 6.01 points in F1.