A class-wise embedding method with class associative and prototype discriminative losses reports state-of-the-art accuracy on six instance-dependent partial label learning benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Mixed Blessing: Class-Wise Embedding guided Instance-Dependent Partial Label Learning
A class-wise embedding method with class associative and prototype discriminative losses reports state-of-the-art accuracy on six instance-dependent partial label learning benchmarks.