DrEM is a dual-side robust ensemble ranking framework that uses a shared logit-space noise model to denoise proxy preferences and stabilize inputs, improving ranking quality under upstream pxtr prediction noise.
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
CONDITIONAL 1representative citing papers
citing papers explorer
-
DrEM: Dual-Side Robust Ensemble Ranking from Noisy User Preference Predictions in Video Recommendation
DrEM is a dual-side robust ensemble ranking framework that uses a shared logit-space noise model to denoise proxy preferences and stabilize inputs, improving ranking quality under upstream pxtr prediction noise.