IEEE Computational Intelligence Magazine - May 2023 - 16

Guest
Editorial
Xingyi Zhang
Anhui University, CHINA
Ran Cheng
Southern University ofScience and Technology, CHINA
Liang Feng
Chongqing University, CHINA
Yaochu Jin
Bielefeld University, GERMANY
Machine Learning Assisted Evolutionary
Multi-Objective Optimization
O
ptimization and learning are
two main paradigms ofartificial
intelligence in addressing complex
real-world problems, with their
respective focuses but frequently enhanced
by each other. Evolutionary multi-objective
optimization (EMO) algorithms are a
family of nature-inspired algorithms
widely used for solving multi-objective
optimization problems (MOPs). Despite
the great success achieved by the existing
EMO algorithms, most ofthese algorithms
have also encountered many challenges in
terms of optimization performance and
efficiency for solving complexMOPs such
as large-scale MOPs, expensive MOPs,
dynamicMOPs, andreal-worldMOPs.
To cope with these challenges, there has
been increasing interest in applying machine
learning (ML) techniques to enhance
the EMO algorithms. Specifically, ML
techniques can be adopted to extract useful
knowledge hidden in the data generated
by EMO algorithms in the search process,
which can be leveraged to assist different
components in EMO algorithms in different
ways, e.g., problem formulation, offspring
generation, fitness evaluation, and/
Digital Object Identifier 10.1109/MCI.2023.3248919
Date ofcurrent version: 6April 2023
or environmental selection. These machine
learning assisted components have
substantially enhanced the ability ofEMO
algorithms in handling complexMOPs.
This special issue has brought
together researchers to report state-ofthe-art
contributions and development
of theories and methodologies on
machine learning techniques for evolutionary
algorithms in addressing complex
MOPs. Following a rigorous peer review
process, three papers have been accepted
for publication in this special issue.
The first paper included in the special
issue is entitled " Adaptive Auxiliary Task
Selection for Multitask-Assisted Constrained
Multi-Objective Optimization "
authored by F. Min et al., which introduces
reinforcement learning techniques for
EMO algorithms to solve constrained
multi-objective optimization problems
(CMOPs). This article proposes a general
multi-tasking framework, where the main
task is for the original CMOP and an arbitrary
numberofauxiliary tasks are designed
based on different strategies for the auxiliary
problems. The reinforcement learning
is used to adaptively choose the most
Corresponding author: Xingyi Zhang (e-mail: 346257113
@qq.com).
16 IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | MAY 2023
suitable auxiliary task in the evolutionary
process. Two popular reinforcement
learning techniques, Q-Learning and deep
Q-Learning, have been considered in the
proposed framework. The empirical results
demonstrated that the reinforcement
learning can enhance the performance of
the proposed multi-tasking framework in
handling CMOPs. The experimental
results also indicate that the proposed
approaches are superior over 11 advanced
algorithms on five benchmark CMOP test
suits and real-world problems.
The second paper entitled " Evolutionary
Multi-Objective Optimization in
Searching for Various Antimicrobial Peptides "
by Y. Liu et al. investigates the use of
EMO algorithms to address the problem
of identifying Antimicrobial peptides
(AMPs) in the sequence space. To ensure
the diversity of the obtained AMPs, a
multi-objective model is formulated by
simultaneously optimizing two objectives,
the antimicrobial activity and diversity
among identified AMPs. An EMO
algorithm is suggested to solve the formulated
MOP, where a deep learning model
constructed based on the bidirectional
encoder representations from transformers
model is employed to predict a peptide's
antimicrobial activity. In the proposed
1556-603X ß 2023 IEEE
IEEE Computational Intelligence Magazine - May 2023

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - May 2023
