IEEE Computational Intelligence Magazine - May 2022 - 67

Chao Wang, Kai Wu, and Jing Liu
Xidian University, CHINA
Frontier
Research
Evolutionary Multitasking AUC Optimization
Abstract
earning to optimize the area
under the receiver operating characteristics
curve (AUC) performance
for imbalanced data has attracted
much attention in recent years.
Although there have been several methods
of AUC optimization, scaling up
AUC optimization is still an open issue
due to its pairwise learning style. Maximizing
AUC in the large-scale dataset
can be considered as a non-convex and
expensive problem. Inspired by the
characteristic of pairwise learning, the
cheap AUC optimization task with a
small-scale dataset sampled from the
large-scale dataset is constructed to promote
the AUC accuracy of the original,
large-scale, and expensive AUC optimization
task. This paper develops an evolutionary
multitasking framework
(termed EMTAUC) to make full use of
information among the constructed
cheap and expensive tasks to obtain
higher performance. In EMTAUC, one
mission is to optimize AUC from the
sampled dataset, and the other is to
maximize AUC from the original dataset.
Moreover, due to the cheap task
containing limited knowledge, a strategy
for dynamically adjusting the data
structure of inexpensive tasks is proposed
to introduce more knowledge
into the multitasking AUC optimization
environment. The performance of the
proposed method is evaluated on a
series of binary classification datasets.
The experimental results demonstrate
that EMTAUC is highly competitive
L
to single task methods and online
methods. Supplementary materials
and source code implementation of
EMTAUC can be accessed at https://
github.com/xiaofangxd/EMTAUC.
I. Introduction
AUC (Area Under ROC Curve) [1] is an
essential measure for characterizing the
performance of machine learning methods
in many real-world applications, such
as ranking and anomaly detection tasks,
and has become one of the most widelyused
performance metrics for coping with
imbalanced data. AUC maximization aims
to obtain a real-valued ranking hypothesis
that places a randomly drawn favorite
instance with a higher decision value than
a randomly drawn annoying instance.
Many efforts have been devoted to
Digital Object Identifier 10.1109/MCI.2022.3155325
Date of current version: 12 April 2022
Corresponding author: Kai Wu (e-mail: kwu@xidian.
edu.cn)
designing efficient batch or online AUC
optimization methods [2], [4], [9], [14].
For batch learning, Joachims [7] firstly
developed a general framework for optimizing
multivariate nonlinear performance
measures such as the AUC and F1.
Linear RankSVM [8] are the most commonly
used AUC maximization methods.
In addition, several efforts optimized the
AUC maximization problem over minibatches
with practical strategies, such as
U-statistics [9], [10], and the adaptive stochastic
gradient method [12]. Unlike the
above methods, there are several evolutionary
algorithms (EA) based methods
to handle this non-convex problem. The
above methods optimize the convex surrogate
of the AUC optimization problem.
In general, EA-based methods
transform the AUC optimization problem
into a bi-objective optimization
problem, which optimizes the True Positive
Rate (TPR) and False Positive Rate
1556-603X/22©2022IEEE
MAY 2022 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE 67
©SHUTTERSTOCK.COM/EAMESBOT
http://www.SHUTTERSTOCK.COM/EAMESBOT http://www.github.com/xiaofangxd/EMTAUC http://www.github.com/xiaofangxd/EMTAUC
IEEE Computational Intelligence Magazine - May 2022

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - May 2022
