VRLS uses entropy regularization of the predictor to improve test-to-train label density ratio estimation, and extends it to multi-node IW-ERM for distributed label shift.
Theorem F .3(Oracle Complexity of Proximal Operator for Composite Optimization)
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
-
Addressing Label Shift in Distributed Learning via Entropy Regularization
VRLS uses entropy regularization of the predictor to improve test-to-train label density ratio estimation, and extends it to multi-node IW-ERM for distributed label shift.