Stochastic Hard Thresholding Algorithms for AUC Maximization

Abstract

In this paper, we aim to develop stochastic hard thresholding algorithms for the important problem of AUC maximization in imbalanced classification. The main challenge is the pairwise loss involved in AUC maximization. We overcome this obstacle by reformulating the U-statistics objective function as an empirical risk minimization (ERM), from which a stochastic hard thresholding algorithm (\texttt{SHT-AUC}) is developed. To our best knowledge, this is the first attempt to provide stochastic hard thresholding algorithms for AUC maximization with a per-iteration cost $Ø(bd)$ where $d$ and $b$ are the dimension of the data and the minibatch size, respectively. We show that the proposed algorithm enjoys the linear convergence rate up to a tolerance error. In particular, we show, if the data is generated from the Gaussian distribution, then its convergence becomes slower as the data gets more imbalanced. We conduct extensive experiments to show the efficiency and effectiveness of the proposed algorithms.

References in corpus (9)

• Structured Variable Selection with Sparsity-Inducing Norms
• A Unified Framework for High-Dimensional Analysis of M-Estimators with Decomposable Regularizers
• Detection of an anomalous cluster in a network
• On Iterative Hard Thresholding Methods for High-dimensional M-Estimation
• Smoothing Multivariate Performance Measures
• On the Generalization Ability of Online Learning Algorithms for Pairwise Loss Functions
• Dual Iterative Hard Thresholding: From Non-convex Sparse Minimization to Non-smooth Concave Maximization
• Sample Efficient Stochastic Gradient Iterative Hard Thresholding Method for Stochastic Sparse Linear Regression with Limited Attribute Observation
• Stochastic Proximal AUC Maximization