RF-DLC trains a particle ensemble with a utility-weighted, class-rebalanced Bayesian objective so long-tailed classifiers can optimize task-specific decision costs rather than plain accuracy.
For example, the method in Section 2 of Elkan (2001) does not consider specific error types during the training phase, and fails to incorporate the utility matrix during testing
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Making Reliable and Flexible Decisions in Long-tailed Classification
RF-DLC trains a particle ensemble with a utility-weighted, class-rebalanced Bayesian objective so long-tailed classifiers can optimize task-specific decision costs rather than plain accuracy.