ABSTRACT
This paper provides a theoretical analysis of rank aggregation consistency under heterogeneous noise models, establishing minimax lower bounds and designing a weighted aggregation estimator that accounts for ranker reliability.
PAPER · PDF
Loading PDF...
Key findings
Derivation of minimax lower bounds for rank aggregation under heterogeneous noise.
Design of a weighted aggregation estimator that achieves minimax optimal rate.
Characterization of phase transitions in consistency under power-law heterogeneity distributions.
Extension of analysis to partial rankings and sparse observation settings.
Limitations & open questions
Theoretical analysis may require empirical validation for practical applications.
Assumptions about noise distributions may not cover all real-world scenarios.