skip to main content

Title: Stochastic Methods for AUC Optimization subject to AUC-based Fairness Constraints
Award ID(s):
Author(s) / Creator(s):
; ;
Date Published:
Journal Name:
The 26th International Conference on Artificial Intelligence and Statistics (AISTATS)
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Chaudhuri, Kamalika and (Ed.)
    In this paper, we propose systematic and efficient gradient-based methods for both one-way and two-way partial AUC (pAUC) maximization that are applicable to deep learning. We propose new formulations of pAUC surrogate objectives by using the distributionally robust optimization (DRO) to define the loss for each individual positive data. We consider two formulations of DRO, one of which is based on conditional-value-at-risk (CVaR) that yields a non-smooth but exact estimator for pAUC, and another one is based on a KL divergence regularized DRO that yields an inexact but smooth (soft) estimator for pAUC. For both one-way and two-way pAUC maximization, we propose two algorithms and prove their convergence for optimizing their two formulations, respectively. Experiments demonstrate the effectiveness of the proposed algorithms for pAUC maximization for deep learning on various datasets. 
    more » « less
  2. null (Ed.)
    A challenge in virology is quantifying relative virulence ( V R ) between two (or more) viruses that exhibit different replication dynamics in a given susceptible host. Host growth curve analysis is often used to mathematically characterize virus–host interactions and to quantify the magnitude of detriment to host due to viral infection. Quantifying V R using canonical parameters, like maximum specific growth rate ( μ max ), can fail to provide reliable information regarding virulence. Although area-under-the-curve (AUC) calculations are more robust, they are sensitive to limit selection. Using empirical data from Sulfolobus Spindle-shaped Virus (SSV) infections, we introduce a novel, simple metric that has proven to be more robust than existing methods for assessing V R . This metric ( I SC ) accurately aligns biological phenomena with quantified metrics to determine V R . It also addresses a gap in virology by permitting comparisons between different non-lytic virus infections or non-lytic versus lytic virus infections on a given host in single-virus/single-host infections. 
    more » « less