The Isotonic Layer unifies calibration and debiasing in recommendation models as a single end-to-end trainable component using learnable context embeddings for piecewise linear adjustments.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Isotonic Layer: A Unified Framework for Recommendation Calibration and Debiasing
The Isotonic Layer unifies calibration and debiasing in recommendation models as a single end-to-end trainable component using learnable context embeddings for piecewise linear adjustments.