Embed Size (px)
Citation preview
Australasian Conference on Artificial Life and
Computational Intelligence (ACALCI 2017)
31 January-2 February 2017, Geelong, VIC, Australia
1
Table of Contents
General Chair and Program Chairs Message ................................................................................................................... 2
Organizing Committee ..................................................................................................................................................... 3
Programme Committee .................................................................................................................................................... 4
Conference location: Deakin Geelong Waterfront Campus ........................................................................................... 5
ACALCI 2017 Program Overview ...................................................................................................................................... 6
Travel to Geelong ............................................................................................................................................................. 7
Train Timetable ................................................................................................................................................................ 8
Keynote Speakers ....................................................................................................................................................... 9-14
Tutorials .................................................................................................................................................................... 15-20
Paper Abstracts ........................................................................................................................................................ 21-27
2
On behalf of the Organizing Committee, it is our great pleasure to welcome you to the Australasian
Conference on Artificial Life and Computational Intelligence (ACALCI 2017), held during January 31 -
February 2, 2017, in Geelong, Australia.
The research areas of artificial life and computational intelligence have grown significantly over recent
years. The breadth is reflected in the papers addressing diverse aspects in the domain, from theoretical
developments to learning, optimization, and applications of such methods to real-world problems.
This conference includes presentations of 32 papers, many of them authored by leading researchers in
the field. After a rigorous evaluation of all 47 submissions by the international Program Committee, 32
manuscripts were selected for single-track oral presentations at ACALCI 2017 (acceptance rate is 68%). All
papers underwent a full peer-review with three to four reviewers per paper.
In addition to the 32 regular paper presentations, ACALCI 2017 will also feature 6 state-of-the-art
plenary talks, 4 informative tutorials, presented by some of the leading experts on the topics.
The ACALCI 2017 international Program Committee consisted of over 63 members from six countries,
based on their affiliation. We would like to thank the members of the international Program Committee,
ACALCI Steering Committee, local Organizing Committee, and other members of the organization team for
their commendable efforts and contributions to the conference.
We would like to acknowledge many supports from RMIT University, The University of Adelaide,
Swinburne University of Technology, and the organizers of the Australian Computer Science Week
(ACSW), who kindly allowed ACALCI 2017 to be co-located with ACSW 2017 at Deakin University,
Geelong.
ACALCI 2017 will provide a stimulating forum for researchers, students, practitioners from around the
world to disseminate their latest research findings and exchange information on emerging research areas in
computational intelligence. The ACALCI 2017 organizers hope you find this conference a pleasant
environment to meet old friends and make new friends. We will make sure that you will have a memorable
experience here!
Xiaodong Li (ACALCI 2017 General Chair)
Markus Wagner, Tim Hendtlass (ACALCI 2017 Program Co-Chairs)
General Chair and Program Chairs Message
3
General Chair: Xiaodong Li (RMIT University)
Program Co-Chairs: Markus Wagner (University of Adelaide)
Tim Hendtlass (Swinburne University of Technology)
Paper & Poster Award Committee Chair: Vic Ciesielski (RMIT University)
Special Session/Workshop Chair: Aldeida Aleti (Monash University)
Treasurer & Registration Chair: Andy Song (RMIT University)
Publicity Chair: Fabio Zambetta (RMIT University)
Kai Qin (RMIT University)
Bing Xue (Victoria University of Wellington)
Local Organizing Committee: Jeffrey Chan (RMIT University)
William Raffe (RMIT University)
Marco Tamassia (RMIT University)
Webmaster: Kelly Gao (University of Adelaide)
Organizing Committee
4
Alan Blair, University of New South Wales
Alan Dorin, Monash University
Aldeida Aleti, Monash University
Andrea Soltoggio, Loughborough University
Andreas Ernst, Monash University
Andrew Lewis, Griffith University
Andy Song, RMIT University
Aneta Neumann, University of Adelaide,
Arindam Dey, University of South Australia
Bing Xue, Victoria University of Wellington
Brad Alexander, University of Adelaide
Brijesh Verma, Central Queensland University
Daniel Le Berre, CNRS - Université d'Artois
Dianhui Wang, La Trobe University
Fabio Zambetta, RMIT University
Frank Neumann, University of Adelaide
Frederic Maire, Queensland University of Technology
Hussein Abbass, University of New South Wales
Ickjai Lee, James Cook University
Inaki Rano, Intelligent Systems Research Centre
Irene Moser, Swinburne University of Technology
Jeff Chan, University of Melbourne
Jianhua Yang, Western Sydney University
Junbin Gao, University of Sydney
Junhua Wu, University of Adelaide
Kai Qin, RMIT University
Kevin Korb, Monash University
Lee Altenberg, Konrad Lorenz Institute for Evolution and Cognition Research
Marc Adam, University of Newcastle, Australia
Marcus Gallagher, University of Queensland
Marcus Randall, Bond University
Markus Wagner, University of Adelaide
Michael Mayo, University of Waikato
Mohamed Abdelrazek, Swinburne University of Technology
Mohammad Reza Bonyadi, University of Adelaide
Muhammad Iqbal, Victoria University of Wellington
Nasser Sabar, RMIT University
Ning Gu, University of South Australia
Oliver Obst, Western Sydney University
Pablo Moscato, University of Newcastle
Paul Kwan, University of New England
Peter Whigham, University of Otago
Programme Committee
Ran Cheng, University of Surrey
Regina Berretta, University of Newcastle
Robert Burdett, Queensland University of Technology
Stephan Chalup, University of Newcastle
Stephen Chen, York University
Tim Hendtlass, Swinburne University
Tom Cai, University of Sydney
Tommaso Urli, CSIRO Data61 / NICTA
Vicky Mak, Deakin University
Wanru Gao, University of Adelaide
William Raffe, RMIT University
Winyu Chinthammit, University of Tasmania
Xiaodong Li, RMIT University
Yi Mei, Victoria University of Wellington
5
Conference location: Deakin Geelong Waterfront Campus
6
Australasian Conference on Artificial Life and Computational Intelligence (ACALCI 2017), Geelong, VIC, Australia
Program Overview
Day 1 (31 Jan 2017)
8.30 am-9.00 am
9.00 am- 10.00 am
10.00 am- 11.20 am
11.20 am- 11.40 am
11.40 am- 1.00 pm
1.00 pm-2.00 pm
2.00 pm- 3.00 pm
3.00 pm- 3.20 pm
3.20 pm- 4.50 pm
4.50 pm- 6.20 pm
7.00 pm onwards
ACALCI opening
Plenary 1: H. Abbass
Chair: X. Li
*18 (Jakomin)
Tea Break
15 (Saleem)
Lunch Break
Plenary 2: Z. Michalewicz
Chair:
T. Hendtlass
Tea Break Tutorial 1:
B. Xue and M. Zhang
Tutorial 2: X. Li
ACSW welcome reception
37 (Haqqani) 4 (Sabar)
19 (Cowley) *14 (Park)
38 (Haqqani) 16 (Masood)
Day 2 (01 Feb 2017)
8.30 am-9.00 am
9.00 am- 10.00 am
10.00 am- 11.20 am
11.20 am- 11.40 am
11.40 am- 1.00 pm
1.00 pm-2.00 pm
2.00 pm- 3.00 pm
3.00 pm- 3.20 pm
3.20 pm- 4.50 pm
4.50 pm- 6.20 pm
ACALCI Conference Dinner
No Session
Plenary 3: K. Smith-Miles
Chair:
M. Wagner
*6 (Qi)
Tea Break
46 (Tan)
Lunch Break
Plenary 4: P. Moscato
Chair:
M. Zhang
Tea Break
7 (Greenwood) Tutorial 3:
A. Trivedi and D. Srinivasan
25 (Habib) 11 (Turky) *10 (Vickers)
35 (Zhang) *5 (Cohen) 29 (Cruz)
47 (Turky) 20 (Gong) 42 (Islam)
Day 3 (02 Feb 2017)
8.30 am-9.00 am
9.00 am- 10.00 am
10.00 am- 11.20 am
11.20 am- 11.40 am
11.40 am- 1.00 pm
1.00 pm-2.00 pm
2.00 pm- 3.00 pm
3.00 pm- 3.20 pm
3.20 pm- 4.50 pm
4.50 pm- 6.20 pm
ACSW Conference Dinner
No Session
Plenary 5: M. Zhang
Chair: X. Li
44 (Yuan)
Tea Break
13 (Sun)
Lunch Break
Plenary 6: D. Green
Chair:
M. Kirley
Tea Break
34 (Fico)
Tutorial 4: K. Qin
27 (Alghamdi) 24 (Wang) 43 (Appana)
41 (Durrani) 3 (Mahbub) *45
(Abdallah)
*8 (Greenwood)
*48 (Trivedi) 33 (Liang)
Conference presentations venue: D2.194 (Building D, level 2, room 194), Deakin Geelong Waterfront Campus: http://www.deakin.edu.au/__data/assets/pdf_file/0019/330364/waterfront.pdf
Conference dinner venue: tba
Plenary
Lectures
Joint ACALCI/ACSW
events
Mathematical
modelling
Tutorials Optimisation Algorithms
and Applications
Notes: 1. The numbers denote paper IDs and last name of the first author. Each paper is allotted 20 mins (15-17 mins presentation + 3-5 mins Q&A/ change-over). 2. For the presentation sessions, the presenters of the paper marked with (*) will chair their respective session.
Prediction Algorithms ALCI
applications
7
Further timetable information:
http://acsw2017.deakin.edu.au/travel/
Travel to Geelong
8
Further timetable information:
https://www.vline.com.au/getattachment/7dcaeccb-adcf-44e1-8f44-e918f34432e0/Geelong-Melbourne-(2)
Train timetable
9
Keynote Speaker 1
Trusted Autonomy: Challenges and Opportunities for Computational Intelligence
Hussein Abbass
University of New South Wales, Australia
Abstract
Trusted Autonomy (TA) is the wider scientific endeavour to establish the groundwork and basic research in
science and engineering required to develop Trusted Autonomous Systems [2]. Autonomous Systems are
leaving the laboratory environment. Will they be used? Will they be safe? Are they ready for a harsh
unpredictable environment? Are they ready to interact with humans? Can they understand humans? Can they
feel humans? Can they trust humans? Will they be trusted?
More questions exist today on autonomous systems than ever before. As the technology becomes
technologically mature, answers to questions on social maturity will become the decisive factor on whether
or not these technologies will be allowed on the streets, in hospitals, schools, or even in the battlespace.
This talk will present on the challenges facing TA, raise questions more than answers, and offer suggestions
for researchers in the field of Computational Intelligence (CI) to work on some of the key challenges in TA;
explaining why it is the right time for CI techniques to showcase their utility in this fast evolving field of
research.
Related publications:
[1] Abbass H.A., Petraki E., Merrick K., Harvey J., and Barlow M. (2016a) Trusted Autonomy and Cognitive Cyber Symbiosis:
Open Challenges, Cognitive Computation, 8(3), 385-408. 10. 1007/s12559-015-9365-5. [ open access ]
[2] Abbass H.A., Leu G., and Merrick K. (2016b) A Review of Theoretical and Practical Challenges of Trusted Autonomy in Big
Data, IEEE Access, 4, 2808 - 2830, doi:10.1109/ACCESS.2016.2571058. [ open access ]
Biography
Hussein Abbass is a Professor of Information Technology at the University of New
South Wales in Canberra (UNSW-Canberra), Australia. He is a fellow of the Australian
Computer Society (FACS), a fellow of the Operational Research Society (FORS,UK); a
fellow of the Australian Institute of Management (FAIM), the President of the
Australian Society for Operations Research, and the Vice-president for Technical
Activities (2016-2017) for the IEEE Computational Intelligence Society. He is an
associate Editor of the IEEE Trans. On Evolutionary Computation, IEEE Trans. on
Cybernetics, IEEE Trans. on Cognitive and Developmental Systems, IEEE
Computational Intelligence Magazine, and four other journals. His current research
contributes to trusted autonomy with an aim to design next generation trusted artificial intelligence systems
that seamlessly integrate humans and machines. His work fuses artificial intelligence, big data, cognitive
science, operations research, and robotics.
Keynote Speakers
10
Keynote Speaker 2
Increasing Sales through Automated Analytics
Zbigniew Michalewicz,
University of Adelaide & Complexica Pty Ltd, Australia
Abstract
The talk is on business applications for transforming data into decisions, based on work done for 3
companies (NuTech Solutions, SolveIT Software, and Complexica) over the last 16 years. A few general
concepts would be discussed, illustrated this by a few examples - from NuTech, from SolveIT, and from
Complexica. The final set of examples would illustrate Complexica’s approach for increasing sales (revenue,
margin, and customer engagement) through automated analysis.
Biography
Zbigniew Michalewicz is the Chief Scientist of Complexica, an Artificial Intelligence
software company that helps large organisations sell more products and services, at a
higher margin, through the use of automated analytics. He is also Emeritus Professor at
the School of Computer Science, University of Adelaide and holds Professor positions at
the Institute of Computer Science, Polish Academy of Sciences, at the Polish-Japanese
Academy of Information Technology, and an honorary Professor position at State Key
Laboratory of Software Engineering of Wuhan University, China. He is also associated
with Structural Complexity Laboratory at Seoul National University, South Korea. In
December 2013 he was awarded (by the President of Poland, Mr. Bronislaw Komorowski) the Order of the
Rebirth of Polish Polonia Restituta - the second highest Polish state decoration civilian (after the Order of
the White Eagle), awarded for outstanding achievements in the field of education, science, sports, culture,
arts, economy, national defence, social activities, the civil service and the development of good relations
with other countries.
For many years his research interests were in the field of evolutionary computation. He published several
books, including a monograph Genetic Algorithms + Data Structures = Evolution Programs (3 editions, a
few translations, over 18,300 citations, source: Google Scholar), and over 250 technical papers in journals
and conference proceedings that are cited widely (over 40,000 citations, source: Google Scholar). He was
one of the editors-in-chief of the Handbook of Evolutionary Computation and the general chairman of the
First IEEE International Conference on Evolutionary Computation held in Orlando, June 1994.
Zbigniew Michalewicz has over 35 years of academic and industry experiences, and possesses expert
knowledge of numerous Artificial Intelligence technologies. He was the co-Founder and Chief Scientist of
NuTech Solutions, which was acquired by Netezza and subsequently by IBM, and the co-Founder and Chief
Scientist of SolveIT Software, which was acquired by Schneider Electric after becoming the 3rd fastest
growing company in Australia. Both companies grew to approximately 200 employees before they were
being acquired. During his time in the corporate world, Professor Michalewicz led numerous large-scale
predictive analytics and optimisation projects for major corporations, including Ford Motor Company, BHP
Billiton, U.S. Department of Defence, and Bank of America. Professor Michalewicz also served as the
Chairman of the Technical Committee on Evolutionary Computation, and later as the Executive Vice
President of IEEE Neural Network Council.
11
Keynote Speaker 3
“We have it all wrong”… so what are you doing to change practice?
Pablo Moscato
The University of Newcastle, Australia
Abstract
Along with many other researchers, I share the view that a systematically coherent research program, in both
theory and applications of algorithms, is definitely needed to accelerate innovation in computing. We
routinely design computational approaches and engage in healthy competitions where the performance of
our methods is tested… but what if “We have it all wrong”? What if we need a paradigmatic change in our
practice for the development and design of computational methods? We may need to enrich our practice
with a new approach.
In fact, John N. Hooker already alerted the computing and mathematical community more than 20 years ago
[Hooker, 1995; Journal of Heuristics]: “Competitive testing tells us which algorithm is faster but not why.”
Hooker argued for a more scientific approach and he proposed the use of ‘controlled experimentation’. This
is common in empirical sciences. “Based on one’s insights into an algorithm”, he said, “one may expect
good performance to depend on a certain problem characteristic”. Then “design a controlled experiment that
checks how the presence or absence of this characteristic affects performance” and, finally, “build an
exploratory mathematical model that captures the insight […] and deduce from its precise consequences that
can be put to the test”.
In this talk, I will address how a new thinking is needed for the development of our field. I will have an with
emphasis in our success on both speeding up solutions for the traveling salesman problem as well as our
success to create very hard instances for the world’s fastest solver.
Biography
Prof. Pablo Moscato is an Australian Research Council Future Fellow and Professor of
Computer Science at The University of Newcastle. At the California Institute of
Technology (1988-89) he developed a methodology called "memetic algorithms"
which is now widely used around the world in Artificial Intelligence, Data Science,
and Business and Consumer Analytics.
Prof. Moscato is Founding Director of the Priority Research Centre for Bioinformatics,
Biomarker Discovery and Information-based Medicine (2007-2015) and the Funding
Director of Newcastle Bioinformatics Initiative (2002-2006) of The University of
Newcastle. His expertise in Data Science was then essential for a large number of applied projects.
He has been working in Applied Mathematics for 30 years, and in heuristic methods for Operations
Research problems since 1985. His work and ideas have been highly influential in a large number of
scientific and technological fields and his manuscripts have been highly cited. The journal "Memetic
Computing" is largely dedicated to a methodology he championed (memetic algorithms). Every 48 hours a
new published paper brings a novel application of these techniques. Due this work and his other
contributions in the areas of classification and machine learning, Prof. Moscato is now well-respected
around the world and he has become one of Australia's most cited computer scientists.
12
Keynote Speaker 4
Instance Spaces for Insightful Performance Evaluation
Kate Smith-Miles
Monash University, Australia
Abstract
Objective assessment of algorithm performance is notoriously difficult, with conclusions often inadvertently
biased towards the chosen test instances. Rather than reporting average performance of algorithms across a
set of chosen instances, we discuss a new methodology to enable the strengths and weaknesses of different
algorithms to be compared across a broader generalised instance space. Initially developed for combinatorial
optimisation, the methodology has recently been extended the domains of continuous optimisation and
machine learning classification. Results will be presented across these problem domains to demonstrate: (i)
how pockets of the instance space can be found where algorithm performance varies significantly from the
average performance of an algorithm; (ii) how the properties of the instances can be used to predict
algorithm performance on previously unseen instances with high accuracy; (iii) how the relative strengths
and weaknesses of each algorithm can be visualized and measured objectively; and (iv) how new test
instances can be generated to fill the instance space and provide insights into algorithmic power.
Biography
Kate Smith-Miles is a Professor in the School of Mathematical Sciences at Monash
University in Australia, where she was Head of School from 2009-2014. She currently
holds a Laureate Fellowship from the Australian Research Council (2014-2019) to
conduct research into new methodologies to gain insights into algorithm strengths and
weaknesses. She is also the inaugural Director of MAXIMA (the Monash Academy
for Cross & Interdisciplinary Mathematical Applications). Kate obtained a B.Sc.(Hons)
in Mathematics and a Ph.D. in Electrical Engineering, both from the University of
Melbourne, Australia. She has published 2 books on neural networks and data mining applications, and over
240 refereed journal and international conference papers in the areas of neural networks, combinatorial
optimization, intelligent systems and data mining. She has supervised to completion 23 PhD students, and
has been awarded over AUD$15 million in competitive grants, including 13 Australian Research Council
grants and industry awards. From December 2016 she will be President of the Australian Mathematical
Society. She was elected Fellow of the Institute of Engineers Australia (FIEAust) in 2006, and Fellow of the
Australian Mathematical Society (FAustMS) in 2008. She was awarded the Australian Mathematical Society
Medal in 2010 for distinguished research. In addition to her academic activities, she also regularly acts as a
consultant to industry in the areas of optimisation, data mining, and intelligent systems.
13
Keynote Speaker 5
Recent Development of Genetic Programming and Applications
Mengjie Zhang
Victoria University of Wellington, New Zealand
Abstract
One of the central challenges of computer science is to use a computer to do what needs to be done without
telling it/knowing the specific process. Genetic programming (GP) addresses this challenge by providing a
method for automatically creating a working computer program from a high-level statement of a specific
task. GP achieves this goal by genetically breeding a population of computer programs using the principles
of Darwinian natural selection and biologically inspired operations. This talk will start with an overview of
GP principles, including representation, operators, search mechanisms and the evolutionary process. The
talk will then discuss the most popular applications of GP with a focus on the evolved "models" and
"generalisation" on symbolic regression and mathematical modelling, classification with unbalanced data,
feature selection and construction, and dynamic job shop scheduling. The talk will end with some interesting
demonstrations and "deep [learning] program structures" in image recognition.
Biography
Mengjie Zhang is currently Professor of Computer Science at Victoria University of
Wellington, where he heads the interdisciplinary Evolutionary Computation Research
Group with 10 staff members and over 20 PhD students. He is a member of the
University Academic Board, a member of the University Postgraduate Scholarships
Committee, a member of the Faculty of Graduate Research Board at the University,
Associate Dean (Research and Innovation) for Faculty of Engineering, and Chair of the
Research Committee for the School of Engineering and Computer Science. His
research is mainly focused on evolutionary computation, particularly genetic
programming, particle swarm optimisation and learning classifier systems with application areas of
computer vision and image processing, multi-objective optimisation, and feature selection and dimension
reduction for classification with high dimensions, transfer learning, classification with missing data, and
scheduling and combinatorial optimisation. Prof Zhang has published over 400 research papers in fully
refereed international journals and conferences in these areas. He has been supervising over 100 research
thesis and project students including over 30 PhD students.
He has been serving as an associated editor or editorial board member for eight international journals
including IEEE Transactions on Evolutionary Computation, the Evolutionary Computation Journal (MIT
Press), Genetic Programming and Evolvable Machines (Springer), Applied Soft Computing, Natural
Computing, and Engineering Applications of Artificial Intelligence, and as a reviewer of over 30
international journals. He has been a major chair for over ten international conferences including IEEE CEC,
GECCO, EvoStar and SEAL. He has also been serving as a steering committee member and a program
committee member for over 80 international conferences including all major conferences in evolutionary
computation. Since 2007, he has been listed as one of the top ten world genetic programming researchers by
the GP bibliography. Prof Zhang is a senior member of IEEE and a member of ACM. He is currently
chairing the IEEE CIS Emergent Technologies Technical Committee consisting of over 40 top CI
researchers from the five continents and 17 task forces. He is the immediate Past Chair for the IEEE CIS
14
Evolutionary Computation Technical Committee and a member of the IEEE CIS Award Committee. He is
also a member of IEEE CIS Intelligent System Applications Technical Committee, a vice-chair of the IEEE
CIS Task Force on Evolutionary Feature Selection and Construction, a vice-chair of the Task Force on
Evolutionary Computer Vision and Image Processing, and the founding chair of the IEEE Computational
Intelligence Chapter in New Zealand.
Keynote Speaker 6
The Network Theory of Complexity
David G. Green
Monash University, Australia
Abstract
Modern society deals increasingly with complex networks. Networks are graphs (nodes linked by arcs or
edges) in which the nodes and/or edges have attributes (e.g., names or sizes). The network theory of
complexity treats properties and processes of complex systems that arise from their underlying networks.
The theory makes strong claims about the universality of networks as well as its relationship to other
theories, such as computational complexity, optimization and Kolmogoroff-Chaitin complexity.
Networks, and network based methods, have gained increasing prominence in many areas of research. This
talk explains the fundamentals of network theory, surveys its applications in various areas of research and
outlines some insights from recent research. It will emphasize important tools, resources and methods
available to researchers.
Theoretical topics covered will include the connectivity avalanche and its implications, network topologies,
motifs, modules, and network metrics. It will also explain several processes in networks that are known to
promote emergent properties, especially feedback, encapsulation and dual-phase evolution.
The talk will include a brief survey of implications of the network properties listed above for understanding
real-world networks. Examples include biological (food webs, genetic regulatory networks), social networks
(group decision-making), technical (computation and communication networks), and socio-technical
(accidents, system failure and supply networks). The talk will conclude by touching on recent developments,
such as interdependent networks.
Biography
David Green is Professor of Information Technology at Monash University. He is
one of Australia’s leading experts on complexity theory. His proof of the
universality of networks showed that networks are inherent in the structure and
behaviour of all complex systems. More recently, he pioneered the theory of dual
phase evolution, a process that explains the way networks contribute to the
emergence of order in many natural and artificial systems. In the course his research
on complexity and evolutionary computing he has investigated problems as diverse as forest ecology,
proteins, geographic information and social networks, and natural computation. He is the author of 9 books,
including Of Ants and Men (2014), Dual-Phase Evolution (2014), Complexity in Landscape Ecology (2006),
and The Serendipity Machine (2004). He is also the author of more than 200 research articles on complexity
theory, evolutionary computing, and multi-agent systems. For further information, please visit his online
profile.
15
Tutorial 1
Advances in Multi-objective Evolutionary Algorithms based on Decomposition
Anupam Trivedi and Dipti Srinivasan, National University of Singapore, Singapore
Abstract
In the last decade, the framework which has attracted the most attention of researchers in the evolutionary
multi-objective optimization community is the decomposition-based framework. Decomposition is a well-
known strategy in traditional multi-objective optimization. However, the decomposition strategy was not
widely employed in evolutionary multiobjective optimization until Zhang and Li proposed multiobjective
evolutionary algorithm based on decomposition (MOEA/D) in 2007. MOEA/D proposed by Zhang and Li
decomposes a multi-objective optimization problem into a number of scalar optimization subproblems, and
optimizes them in a collaborative manner using an evolutionary algorithm. Each subproblem is optimized by
utilizing the information mainly from its several neighbouring subproblems. Since the proposition of
MOEA/D in 2007, several studies have been conducted in the literature to: a) overcome the limitations in the
design components of the original MOEA/D, b) improve the performance of MOEA/D, c) present novel
decomposition-based MOEAs, and d) adapt decomposition-based MOEAs for different type of problems.
Investigations on the decomposition-based framework have been undertaken in several directions, including
development of novel weight vector generation methods, use of new decomposition approaches, efficient
allocation of computational resources, modifications in the reproduction operators, mating selection and
replacement mechanism, hybridizing decomposition- and dominance-based approaches, etc. Furthermore,
several attempts have been made at extending the decomposition-based framework to constrained multi-
objective optimization, many-objective optimization, and incorporate the preference of decision makers.
This tutorial will present a comprehensive survey of the decomposition-based MOEAs proposed in the last
decade. We will highlight the strengths and weakness of the different decomposition-based MOEAs
presented in the literature and present interesting directions for future research. Finally, we will present our
newly designed MOEA/D variant to incorporate the a priori preferences of the decision maker, namely
pMOEA/D. We will present experimental results on benchmark functions from different test suites such as
ZDT, DTLZ, and UF, to demonstrate the effectiveness of the proposed algorithm.
This tutorial should be of interest to both new beginners and experienced researchers in the area of
multiobjective optimization. The tutorial will provide a unique opportunity to showcase the latest
development on this hot research topic to the EC research community. We expect the tutorial will last about
110 minutes.
Biography
Anupam Trivedi received his received the Dual degree (integrated Bachelor’s and Master’s) in Civil
Engineering from the Indian Institute of Technology (IIT) Bombay, Mumbai, India, in 2009, and the Ph.D.
degree in Electrical & Computer engineering from the National University of Singapore, Singapore, in 2015.
Currently, he is a Research Fellow at the Department of Electrical & Computer Engineering, National
University of Singapore, Singapore. His research interests include evolutionary computation, multiobjective
optimization, and power systems.
Tutorials
16
Dipti Srinivasan received the Ph.D. degree in engineering from the National University of Singapore,
Singapore, in 1994. She worked as a Postdoctoral Researcher with the University of California, Berkeley,
CA, USA, from 1994 to 1995, before joining the National University of Singapore, where she is currently an
Associate Professor with the Department of Electrical and Computer Engineering. Her research interests
include evolutionary computation, neural networks, multiobjective optimization, and power systems. She is
currently serving as an Associate Editor of IEEE Transactions on Evolutionary Computation, IEEE
Transactions on Neural Networks and Learning Systems, IEEE Computational Intelligence magazine, IEEE
Transactions on Intelligent Transportation Systems, and IEEE Transactions on Sustainable Energy.
Tutorial 2
Evolutionary Feature Selection and Feature Construction for Dimensionality Reduction
Bing Xue, Mengjie Zhang, Victoria University of Wellington, New Zealand
Abstract
In data mining and machine learning, many real-world problems such as bio-data classification and
biomarker detection, image analysis, text mining often involve a large number of features/attributes.
However, not all the features are essential since many of them are redundant or even irrelevant, and the
useful features are typically not equally important. Using all the features for classification or other data
mining tasks typically does not produce good results due to the big dimensionality and the large search
space. This problem can be solved by feature selection to select a small subset of original (relevant) features
or feature construction to create a smaller set of high-level features using the original low-level features.
Feature selection and construction are very challenging tasks due to the large search space and feature
interaction problems. Exhaustive search for the best feature subset of a given dataset is practically
impossible in most situations. A variety of heuristic search techniques have been applied to feature selection
and construction, but most of the existing methods still suffer from stagnation in local optima and/or high
computational cost. Due to the global search potential and heuristic guidelines, evolutionary computation
techniques such as genetic algorithms, genetic programming, particle swarm optimisation, ant colony
optimisation, differential evolution and evolutionary multi-objective optimisation have been recently used
for feature selection and construction for dimensionality reduction, and achieved great success. Many of
these methods only select/construct a small number of important features, produce higher accuracy, and
generated small models that are efficient on unseen data. Evolutionary computation techniques have now
become an important means for handle big dimensionality and feature selection and construction.
The tutorial will introduce the general framework within which evolutionary feature selection and
construction can be studied and applied, sketching a schematic taxonomy of the field and providing
examples of successful real-world applications. The application areas to be covered will include bio-data
classification and biomarker detection, image analysis and object recognition and pattern classification,
symbolic regression, network security and intrusion detection, and text mining. EC techniques to be covered
will include genetic algorithms, genetic programming, particle swarm optimisation, differential evolution,
ant colony optimisation, artificial bee colony optimisation, and evolutionary multi-objective optimisation.
We will show how such evolutionary computation techniques can be effectively applied to feature
selection/construction and dimensionality reduction and provide promising results.
17
This tutorial should be of interest to both new beginners and experienced researchers in the area of
evolutionary feature selection and feature construction. The tutorial will provide a unique opportunity to
showcase the latest development on this hot research topic to the research community. We expect the
tutorial will last about 110 minutes.
Biography
Bing Xue is currently a lecturer and co-leader of the Evolutionary Computation Research Group, School of
Engineering and Computer Science at Victoria University of Wellington, and leading the strategic research
direction on evolutionary feature selection and construction. Her research focuses mainly on evolutionary
computation, pattern recognition, feature selection, feature construction, multi-objective optimisation, data
mining and machine learning.
She has over 70 papers published in fully referred international journals and conferences and most of them
are on evolutionary feature selection and construction. She is currently co-supervising over 10 PhD and
Master’s students and visiting scholars, and over 10 Honours and summer research projects.
Dr Xue is the main Chair of IEEE Symposium on Computational Intelligence in Feature Analysis, Selection,
and Learning in Image and Pattern Recognition (FASLIP) in IEEE Symposium Series on Computational
Intelligence (IEEE SSCI 2016), a program co-chair of the 7th International Conference on Soft Computing
and Pattern Recognition (SoCPaR2015), Publicity Chair for the Australian Conference on Artificial Lift and
Computational Intelligence (ACALCI 2017), Special Session co-Chair for The 20th Asia-Pacific
Symposium on Intelligent and Evolutionary Systems (IES2016), Special Session Co-chair on Evolutionary
Feature Reduction in the international conference on Simulated Evolution And Learning (SEAL 2014), and
the main organiser of the special session on Evolutionary Feature Selection and Construction in IEEE
Congress on Evolutionary Computation (CEC) 2015 and 2016. She is a member of Editorial Board for
Applied Soft Computing (journal), International Journal of Computer Information Systems and Industrial
Management Applications and International Journal of Swarm Intelligence Research, and also a Guest
Editor for the Special Issue on Evolutionary Feature Reduction and Machine Learning for the Springer
Journal of Soft Computing. Dr Xue is serving as a reviewer of over 10 international journals including IEEE
Transactions on Evolutionary Computation, IEEE Transaction on Cybernetics and Information Sciences.
She is a program committee member for many international conferences including Genetic and Evolutionary
Computation Conference (GECCO), European Joint Conference on Evolutionary Computation (EvoStar --
EuroGP, EvoCOP and EvoApplications), IEEE Congress on Evolutionary Computation (CEC), International
Joint Conference on Artificial Intelligence (IJCAI), Pacific-Asia Conference on Knowledge Discovery and
Data Mining (PAKDD), and International Conference on Simulated Evolution and Learning (SEAL).
Dr Xue is currently the Chair of the IEEE Task Force on Evolutionary Feature Selection and Construction,
consisting of over 20 members for the five continents working in this area. She is also serving as the
Director of Women in Engineering for the IEEE New Zealand Central Section and the Secretary of the IEEE
Chapter on Computational Intelligence in that Section.
Mengjie Zhang is currently Professor of Computer Science at Victoria University of Wellington, where he
heads the interdisciplinary Evolutionary Computation Research Group. He is a member of the University
Academic Board, a member of the University Postgraduate Scholarships Committee, a member of the
Faculty of Graduate Research Board at the University, Associate Dean (Research and Innovation) in the
Faculty of Engineering, and Chair of the Research Committee of the Faculty of Engineering and School of
Engineering and Computer Science.
18
His research is mainly focused on evolutionary computation, particularly genetic programming, particle
swarm optimisation and learning classifier systems with application areas of feature selection/construction
and dimensionality reduction, computer vision and image processing, job shop scheduling, multi-objective
optimisation, and classification with unbalanced and missing data. He is also interested in data mining,
machine learning, and web information extraction. Prof Zhang has published over 400 research papers in
refereed international journals and conferences in these areas.
He has been serving as an associated editor or editorial board member for seven international journals
including IEEE Transactions on Evolutionary Computation, the Evolutionary Computation Journal (MIT
Press), Genetic Programming and Evolvable Machines (Springer), Applied Soft Computing, IEEE
Transactions on Emergent Topics in Computational Intelligence, Natural Computing, and Engineering
Applications of Artificial Intelligence, and as a reviewer of over 30 international journals. He has been
involving major EC conferences such as GECCO, IEEE CEC, EvoStar, IEEE SSCI and SEAL as a Chair.
He has also been serving as a steering committee member and a program committee member for over 100
international conferences including all major conferences in evolutionary computation. Since 2007, he has
been listed as one of the top ten world genetic programming researchers by the GP bibliography.
Prof Zhang is the Chair of the IEEE Emergent Technologies Technical Committee, the immediate Past Chair
of the IEEE CIS Evolutionary Computation Technical Committee, a vice-chair of the IEEE CIS Task Force
on Evolutionary Feature Selection and Construction, a vice-chair of the IEEE CIS Task Force on
Evolutionary Computer Vision and Image Processing, and the founding chair of the IEEE Computational
Intelligence Chapter in New Zealand.
Tutorial 3
Recent Advances in Multimodal Optimization using Niching Methods
Xiaodong Li, RMIT University, Australia
Abstract
This talk provides an update on the recent development of multimodal optimization methods, also
commonly referred to as niching methods, in the area of evolutionary computation. Population-based meta-
heuristic algorithms such as Evolutionary Algorithms (EAs) in their original forms are usually designed for
locating a single global solution. These algorithms typically converge to a single solution because of the
global selection scheme used. Nevertheless, many real-world problems are “multimodal” by nature, i.e.,
multiple satisfactory solutions exist. It may be desirable to locate many such satisfactory solutions so that a
decision maker can choose one that is most proper in his/her problem domain. A niching method can be
incorporated into a standard EA to promote and maintain formation of multiple stable subpopulations within
a single population, with an aim to locate multiple globally optimal or suboptimal solutions. Many niching
methods have been developed in the past, including crowding, fitness sharing, restricted tournament
selection, clearing, speciation, etc. In more recent times, niching methods have also been developed for other
meta-heuristic algorithms such as Particle Swarm Optimization (PSO) and Differential Evolution (DE).
Currently niching methods are experiencing a revival, attracting researchers from across a wide range of
research fields. This talk will first provide information on the origin of niching methods, motivation on why
research on niching is important, and its practical relevance to real-world problem solving. It will revisit
some of the most classic niching methods, before showing some recent development on niching methods,
19
harnessing the unique characteristics of PSO and DE. I will then go on to show how niching methods can be
employed to enhance performance in dynamic optimization and multiobjective optimization. Finally I will
discuss the top entries from the IEEE CEC’2015 niching methods competition.
Related publications:
Parrott, D. and Li, X. (2006), “Locating and Tracking Multiple Dynamic Optima by a Particle Swarm Model Using
Speciation”, IEEE Transactions on Evolutionary Computation, 10(4):440-458, August 2006.
Li, X., Branke, J. and Blackwell, T. (2006), “Particle Swarm with Speciation and Adaptation in a Dynamic
Environment”, in Proceeding of Genetic and Evolutionary Computation Conference 2006 (GECCO'06), eds. M.
Keijzer, et al., pp.51 - 58, ACM Press.
Li, X. (2010), “Niching without Niching Parameters: Particle Swarm Optimization Using a Ring Topology”, IEEE
Transactions on Evolutionary Computation, 14 (1): 150-169, February 2010.
Epitropakis, M., Li, X. and Burke, E. (2013), “A Dynamic Archive Niching Differential Evolution algorithm for
Multimodal Optimization”, in Proceedings of Congress of Evolutionary Computation (CEC 2013), IEEE, pp.79 - 86.
Li, X., Engelbrecht, A. and Epitropakis, M.G. (2013), “Benchmark Functions for CEC'2013 Special Session and
Competition on Niching Methods for Multimodal Function Optimization”, Technical Report, Evolutionary
Computation and Machine Learning Group, RMIT University, Australia, 2013.
Biography
Xiaodong Li received his B.Sc. degree from Xidian University, Xi'an, China, and Ph.D. degree in
information science from University of Otago, Dunedin, New Zealand, respectively. Currently, he is an
Associate Professor at the School of Computer Science and Information Technology, RMIT University,
Melbourne, Australia. His research interests include evolutionary computation, neural networks, complex
systems, multiobjective optimization, and swarm intelligence. He serves as an Associate Editor of the IEEE
Transactions on Evolutionary Computation, the journal of Swarm Intelligence (Springer), and International
Journal of Swarm Intelligence Research. He is a founding member and currently a Vice-chair of the
following three IEEE CIS Task Forces: Swarm Intelligence, Large Scale Global Optimization, and
Multimodal Optimization. He was the General Chair of SEAL'08, a Program Co-Chair AI'09, and a Program
Co-Chair for IEEE CEC’2012. He is the recipient of 2013 SIGEVO Impact Award, and 2017 IEEE CIS
"IEEE Transactions on Evolutionary Computation Outstanding Paper Award". For further information,
please visit his website.
20
Tutorial 4
Data-Driven Evolutionary Optimisation
Kai Qin, RMIT University, Australia
Abstract
Optimisation aims to find the best solution from numerous feasible ones, which is demanded in nearly every
field. Evolutionary optimisation represents a family of optimisation techniques based on Darwinian
principles, characterised by a population of candidate solutions evolved via nature-inspired operations to
search for the optimum. Intrinsically, it belongs to a generate-and-test problem solver which will produce a
large volume of “data” (e.g., candidate solutions and their related information) as search progresses. Such
"data" may contain prolific information about the behaviours of optimisation methods and the properties of
optimisation tasks.
In the past few decades, a lot of efforts had been made to enhance evolutionary optimisation techniques via
exploiting (e.g., using data analytics techniques) the “data” generated in the course of search. However,
modern optimisation problems, featured with the fast-growing scale, complexity and uncertainty, can
seldom be tackled by simply hybridising evolutionary optimisation with some off-the-shelf data analytics
tools, and therefore call for an in-depth investigation on how to leverage the “data” generated during search
to facilitate optimisation.
This tutorial aims to introduce a unified perspective on evolutionary optimisation techniques that adopts data
analytics as an indispensable component, describe how to identify and address various data analytics tasks in
the search process, and discuss an emerging research trend which makes use of search experience gained by
solving some problems to facilitate solving other problems via knowledge transfer. The audience is expected
to get to know the fundamentals and recent developments in data-driven evolutionary optimisation, and be
inspired to employ such techniques to deal with their encountered optimisation problems.
Biography
Dr. Kai Qin received his BEng degree from Southeast University, China, in 2001, and his PhD from
Nanyang Technological University, Singapore, in 2007. He is now a lecturer in the School of Science at
RMIT University. His major research interests include evolutionary computation, machine learning, image
processing, GPU computing, and services computing. He has published 70+ papers and received two best
paper awards. Two of his co-authored papers are the 1st and 4th most cited papers (Web of Science) in IEEE
Transactions on Evolutionary Computation over the last 10 years. He is currently chairing the IEEE
Computational Intelligence Society task force on Collaborative Learning and Optimisation, promoting
research on synergising machine learning and intelligent optimisation to resolve challenging real-world
problems which involve learning and optimisation as indispensable and interwoven tasks.
21
Paper#3: Md. Shahriar Mahbub, Markus Wagner and Luigi Crema.
Multi-Objective Optimisation with Multiple Preferred Regions
Abstract: The typical goal in multi-objective optimization is to find a set of good and well-distributed solutions. It has become
popular to focus on specific regions of the objective space, e.g., due to market demands or personal preferences. In the past, a
range of different approaches has been proposed to consider preferences for regions, including reference points and weights.
While the former technique requires knowledge over the true set of trade-offs (and a notion of "closeness") in order to perform
well, it is not trivial to encode a non-standard preference for the latter. With this article, we contribute to the set of algorithms that
consider preferences. In particular, we propose the easy-to-use concept of "preferred regions" that can be used by laypeople, we
explain algorithmic modifications of NSGAII and AGE, and we validate their effectiveness on benchmark problems and on a real-
world problem.
Paper#4: Nasser R. Sabar and Aldeida Aleti.
An Adaptive Memetic Algorithm for the Architecture Optimisation Problem
Abstract: Architecture design is one of the most important steps in software development, since design decisions affect the
quality of the final system (e.g. reliability and performance). Due to the ever-growing complexity and size of software systems,
deciding on the best design is a computationally intensive and complex task. This issue has been tackled by using optimisation
method, such as local search and genetic algorithms. Genetic algorithms work well in rugged fitness landscapes, whereas local
search methods are successful when the search space is smooth. The strengths of these two algorithms have been combined to
create memetic algorithms, which have shown to be more efficient than genetic algorithms and local search on their own. Two
major points of concern with memetic algorithms are (i) the computational cost of calling the local search method at every
iteration, which evaluates all neighbours, and (ii) the likelihood of loosing the exploration capacity because of the `exploitative'
nature of local search. To address these two issues, this work uses an adaptive scheme to control the local search application. The
utilised scheme takes into account the diversity of the current population. Based on the diversity indicator, it decides whether to
call local search or not. Experiments were conducted on the component deployment problem to evaluates the effectiveness of the
proposed algorithm with and without the adaptive local search algorithm.
Paper#5: Dror Cohen, Antonio Gomez, Dhananjay Thiruvady and Andreas Ernst.
Resource Constrained Job Scheduling with Parallel Constraint-based ACO
Abstract: Hybrid methods are highly effective means of solving combinatorial optimization problems and have become
increasingly popular. In particular, integrations of exact and incomplete methods have proved to be effective where the hybrid
takes advantage of the relative performance of each individual method. However, these methods often require significant run-
times to determine good feasible solutions. One way of reducing run-times is to parallelize these algorithms. For large NP--hard
problems, parallelization must be done with care, since changes to the algorithm can affect its performance in unpredictable ways.
In this paper we develop two parallel variants of constraint-based ACO and test them on a problem motivated in the Australian
mining industry. We demonstrate that parallelization significantly reduces run times with each parallel variant providing
advantages with respect to feasibility or solution quality.
Paper#6: Yutao Qi, Haodong Guo and Xiaodong Li.
Extending the Delaunay Triangulation Based Density Measurement to Many-objective Optimization
Abstract: This paper investigates the scalability of the Delaunay triangulation (DT) based diversity preservation technique for
solving many-objective optimization problems (MaOPs). Following the NSGA-II algorithm, the proposed optimizer with DT
based density measurement (NSGAII-DT) determines the density of individuals according to the DT mesh built on the population
in the objective space. To reduce the computing time, the population is projected onto a plane before build-ing the DT mesh.
Experimental results show that NSGA-II-DT outperforms NSGA-II on WFG problems with 4, 5 and 6 objectives. Two projection
strategies using a unit plane and a least-squares plane in the objective space are investigated and compared. Our results also show
that the former is more effective than the latter.
Paper#7: Garrison Greenwood, Hussein Abbass and Eleni Petraki.
Emotion, Trustworthiness and Altruistic Punishment in a Tragedy of the Commons Social Dilemma
Abstract: Social dilemmas require individuals to tradeoff self interests against group interests. Considerable research effort has
attempted to identify conditions that promote cooperation in these social dilemmas. It has previously been shown altruistic
punishment can help promote cooperation but the mechanisms that make it work are not well understood. We have designed a
multi-agent system to investigate altruistic punishment in tragedy of the commons social dilemmas. Players develop emotional
responses as they interact with others. A zero order Seguno fuzzy system is used to model the player emotional responses. Players
change strategies when their emotional level exceeds a personal emotional threshold. Trustworthiness of how other players will
ACALCI 2017 Accepted Papers with Abstracts
22
act in the future helps choose the new strategy. Our results show how strategies evolve in a finite population match predictions
made using discrete replicator equations.
Paper#8: Garry Greenwood, Richard Tymerski and Devin Sills.
Equity Option Strategy Discovery and Optimization Using a Memetic Algorithm
Abstract: Options in finance are becoming an increasingly popular investment instrument. Good returns, however, do depend on
finding the right strategy for trading and risk management. In this paper we describe a memetic algorithm designed to discover
and optimize multi-leg option strategies for the S&P500 index. Strategies comprising one up to six option legs are examined. The
fitness function is specifically designed to maximize profitability while seeking a certain trade success percentage and equity
drawdown limit. Using historical option data from 2005 to 2016, our memetic algorithm discovered a four-leg option strategy that
offers optimum performance.
Paper#10: Darwin Vickers, Jacob Soderlund and Alan Blair.
Co-Evolving Line Drawings with Hierarchical Evolution
Abstract: We use an approach inspired by biological coevolution to generate complex line drawings without human guidance.
Artificial artists and critics work against each other in an iterative competitive framework, forcing each to become increasingly
sophisticated to outplay the other. Both the artists and critics are implemented in HERCL, a framework combining linear and
stack-based Genetic Programming, which is well suited to coevolution because the number of competing agents is kept small
while still preserving diversity. The aesthetic quality of the resulting images lies in the ability of the evolved HERCL programs,
making judicious use of register adjustments and loops, to produce repeated substructures with subtle variations, in the spirit of
low-complexity art.
Paper#11: Ayad Turky, Irene Moser and Aldeida Aleti.
An Iterated Local Search with Guided Perturbation for the Heterogeneous Fleet Vehicle Routing Problem with Time
Windows and Three-Dimensional Loading Constraints
Abstract: An Australian company is faced with the logistics problem of distributing small quantities of fibre boards to hundreds
of customers every day. The resulting Heterogeneous Fleet Vehicle Routing Problem with Time Windows and Three-Dimensional
Loading Constraints has to be solved within a single hour, hence the use of a heuristic instead of an exact method. In previous
work, the loading was performed after optimising the routes, which in some cases generated infeasible solutions in need of a
repair mechanism. In this work, the feasibility of the loading constrains is maintained during the route optimisation. Iterated Local
Search algorithm has proved very effective at solving vehicle routing problems. Its success is mainly due to its biased sampling of
local optima. However, its performance heavily depends on the perturbation procedure. We trialled different perturbation
procedures where the first one perturbs the give solution by moving customers incurring the highest cost on the objective function,
whilst the second one moves customers that have a smaller number of deliveries. Our industry partner provided six sets of daily
orders which have varied characteristics in terms of the number of customers, customer distribution, number of fibre boards and
fibre boards' sizes. Our investigations show that an instance becomes more constrained when the customer order contains many
different board sizes, which makes it harder to find feasible solutions. The results show that the proposed perturbation procedures
significantly enhances the performance of iterated local search specifically on such constrained problems.
Paper#13: Yuan Sun, Michael Kirley and Saman Halgamuge.
A Memetic Cooperative Co-evolution Model for Large Scale Continuous Optimization
Abstract: Cooperative co-evolution (CC) is a framework that can be used to `scale up' EAs to solve high dimensional
optimization problems. This approach employs a divide and conquer strategy, which decomposes a high dimensional problem into
sub-components that are optimized separately. However, the traditional CC framework typically employs only one EA to solve all
the sub-components, which may be ineffective. In this paper, we propose a new memetic cooperative co-evolution (MCC)
framework which divides a high dimensional problem into several separable and non-separable sub-components based on the
underlying structure of variable interactions. Then, different local search methods are employed to enhance the search of an EA to
solve the separable and non-separable sub-components. The proposed MCC model was evaluated on two benchmark sets with 35
benchmark problems. The experimental results confirmed the effectiveness of our proposed model, when compared against two
traditional CC algorithms and a state-of-the-art memetic algorithm.
Paper#14: John Park, Yi Mei, Su Nguyen, Gang Chen and Mengjie Zhang.
Investigating the Generality of Genetic Programming based Hyper-heuristic Approach to Dynamic Job Shop Scheduling
with Machine Breakdown
Abstract: Dynamic job shop scheduling (DJSS) problems are combinatorial optimisation problems that have been extensively
studied in the literature due their difficulty and their applicability to real-world manufacturing systems, e.g., car manufacturing
systems. In a DJSS problem instance, jobs arrive on the shop floor to be processed on a specific sequence of machines on the shop
floor. DJSS problems with unforeseen events such as dynamic job arrivals and machine breakdowns are examples of DJSS
23
problems that have been studied frequently in the literature. In particular, many researchers have proposed genetic programming
based hyper-heuristic (GP-HH) approaches to evolve high quality dispatching rules for DJSS problems with dynamic job arrivals,
outperforming good man-made rules for the problems. However, no GP-HH approaches have been proposed for DJSS problems
with dynamic job arrivals and machine breakdowns, and it is not know how well GP generalises over both DJSS problem
instances with no machine breakdown to problem instances with machine breakdown. Therefore, this paper investigates the
generality of GP for DJSS problem with dynamic job arrivals and machine breakdowns and analyses the structure of the rules
evolved for specific machine breakdown scenarios. The results show that performance and the distributions of the terminals for
the evolved rules is sensitive to the frequency of machine breakdowns in the training instances used to evolve the rules.
Paper#15: Sobia Saleem and Marcus Gallagher.
Exploratory Analysis of Clustering Problems Using a Comparison of Particle Swarm Optimization and Differential
Evolution
Abstract: The size, scope and variety of the experimental analyses of metaheuristics has increased in recent years, aiming to
develop new procedures and techniques to improve our understanding of optimization algorithms and problems. In this paper, we
compare particle swarm optimization and differential evolution on a set of real-world clustering problems. Generally,
experimental comparisons focus on presenting a statistical summary of algorithm performance, however this hides valuable
information about the algorithm behaviour on the problems in question. We take an exploratory approach, focussing on extracting
deeper insights and understanding from the experimental results data. We make progress on understanding the fitness landscapes
of the set of clustering problems, as well as analysing current and previous experimental results for algorithms applied to these
problems. Consequently, the paper makes two contributions: (a) Advancing our understanding of what factors make this set of
problem instances easy or hard for given algorithms; (b) Demonstrating the need to be careful in experimental evaluations and that
better insights can be obtained with exploratory analysis.
Paper#16: Atiya Masood, Gang Chen, Yi Mei and Mengjie Zhang.
A PSO-based Reference Point Adaption Method for Genetic Programming Hyper-heuristic in Many-Objective Job Shop
Scheduling
Abstract: Job Shop Scheduling is one of the most important combinatorial optimisation problems in practice. It usually contains
many (four or more) potentially conflicting objectives such as makespan and mean weighted tardiness. On the other hand,
evolving dispatching rules using genetic programming has demonstrated to be a promising approach to solving job shop
scheduling due to its flexibility and scalability. In this paper, we aim to solve many-objective job shop scheduling with genetic
programming and NSGA-III. However, NSGA-III is originally designed to work with uniformly distributed reference points
which do not match well with the discrete and non-uniform Pareto front in job shop scheduling problems, resulting in many
useless points during evolution.These useless points can significantly affect the performance of NSGA-III and genetic
programming. To address this issue and inspired by particle swarm optimisation, a new reference point adaptation mechanism has
been proposed in this paper. Experiment results on many-objective benchmark job shop scheduling instances clearly show that
prominent improvement in performance can be achieved upon using our reference point adaptation mechanism in NSGA-III and
genetic programming.
Paper#18: Martin Jakomin and Zoran Bosnić.
Reliability estimation of individual multi-target regression predictions
Abstract: To estimate the quality of the induced predictive model we generally use measures of averaged prediction accuracy,
such as the relative mean squared error on test data. Such evaluation fails to provide local information about reliability of
individual predictions, which can be important in risk-sensitive fields (medicine, finance, industry etc.). Related work presented
several ways for computing individual prediction reliability estimates for single-target regression models, but has not considered
their use with multi-target regression models that predict a vector of independent target variables. In this paper we adapt the
existing single-target reliability estimates to multi-target models. We approach this in two ways: by aggregating reliability
estimates for individual target components, and by generalizing the existing reliability estimates to higher number of dimensions.
The results revealed favorable performance of the reliability estimates that are based on bagging variance and local cross-
validation approaches. The results are consistent with the related work in single-target reliability estimates and provide a support
for multi-target decision making.
Paper#19: Benjamin Cowley and John Thornton.
Feedback Modulated Attention Within a Predictive Framework
Abstract: Attention is both ubiquitous throughout and key to our cognitive experience. It has been shown to filter out mundane
stimuli, while simultaneously communicating specific stimuli from the lowest levels of perception through to the highest levels of
cognition. In this paper we present a connectionist system with mechanisms that produce both exogenous (bottom-up) and
endogenous (top-down) attention. The foundational algorithm of our system is the Temporal Pooler (TP), a neocortically inspired
algorithm that learns and predicts temporal sequences. We make a number of modifications to the Temporal Pooler and place it in
a framework which is inspired by predictive coding. We use a novel technique in which feedback connections elicit endogenous
24
attention by disrupting the learned representations of attended sequences. Our experiments show that this approach successfully
filters attended stimuli and suppresses unattended stimuli.
Paper#20: Wenyin Gong.
Optimal power allocation of wireless sensor networks with multi-operator based constrained differential evolution
Abstract: Optimal power allocation (OPA) is considered to be one of the key issues in designing a wireless sensor network
(WSN). Generally, the OPA in WSN can be formulated as a numerical optimization problem with constraints. Differential
evolution (DE) is a powerful evolutionary algorithm for numerical, however, the success of DE in solving a specific problem
crucially depends on appropriately choosing suitable mutation strategy and its associated control parameter values. Meanwhile,
there is no single parameter setting and strategy that is able to consistently obtain the best results for the OPA with different
number of sensor nodes. Based on the above considerations, in this paper, a multi-operator based constrained differential
evolution is proposed, where probability matching and constrained credit assignment techniques are used so as to adaptively select
the most suitable strategy in different phase of the search process for the OPA. Additionally, the parameter adaptation technique is
used to avoid the fine-tuning of DE parameters for different problems. The proposed algorithm has been evaluated in several OPA
with different number of sensor nodes, and its performance is compared with single-strategy based DE variants and other methods.
Experimental results indicate that the proposed algorithm is able to provide better results than the compared methods.
Paper#24: Erli Wang, Hanna Kurniawati and Dirk Kroese.
CEMAB: A Cross-Entropy-based Method for Large-Scale Multi-Armed Bandits
Abstract: The multi-armed bandit (MAB) problem is an important model for studying the exploration--exploitation tradeoff in
sequential decision making. In this problem, a gambler has to repeatedly choose between a number of slot machine arms to
maximize the total payout, where the total number of plays is fixed. Although many methods have been proposed to solve the
MAB problem, most have been designed for problems with a small number of arms. To ensure convergence to the optimal arm,
many of these methods, including state-of-the-art methods such as UCB, require sweeping over the entire set of arms. As a result,
such methods perform poorly in problems with a large number of arms. This paper proposes a method for solving such large-scale
MAB problems. The method, called Cross-Entropy-based Multi Armed Bandit (CEMAB), uses the Cross-Entropy method as a
noisy optimizer to find the optimal arm with as little cost as possible. Experimental results indicate that CEMAB outperforms
state-of-the-art methods for solving MABs with a large number of arms.
Paper#25: Ahsanul Habib, Hemant Kumar Singh and Tapabrata Ray.
A Batch Infill Strategy for Computationally Expensive Optimization Problems
Abstract: Efficient Global Optimization (EGO) is a well established iterative scheme for solving computationally expensive
optimization problems. EGO relies on an underlying Kriging model and maximizes the expected improvement (EI) function to
obtain an infill (sampling) location. The Kriging model is in turn updated with this new truly evaluated solution and the process
continues until the termination (usually the computational budget) condition is met. The serial nature of the process limits its
efficiency for applications where a batch of solutions can be evaluated at the same cost as a single solution. Examples of such
cases include (a) batch of physical experiments for drug design, material synthesis or (b) computational analysis that can be
executed in parallel. In this paper we present a multi-objective formulation to deal with such classes of problems including
constrained problems, where instead of a single solution, a batch of solutions are identified and evaluated concurrently. The
strategies use different objectives depending on the history of evaluated solutions (all feasible solutions, at least one feasible
solution or all infeasible solutions). The performance of the strategy is analyzed using well studied benchmarking problems
covering the above problem classes from the literature and compared with contemporary MO formulation based approaches.
Studies are also conducted for multiple batch sizes for completeness.
Paper#27: Mahfouth Alghamdi and Haifeng Shen.
Automatic Clustering and Summarisation of Microblogs: A Multi-Subtopic Phrase Reinforcement Algorithm
Abstract: There is a phenomenal growth of microblogging-based social communication services and subscriptions in recent years.
Through these services, users publish a large number of posts within a short period time, making it extremely hard for readers to
keep track of a trending topic. A solution to this issue is text summarisation, which can generate a short summary of a trending
topic from multiple posts. Most of the existing summarisation algorithms were proposed for long documents and do not work well
for short microblogging posts. The PR (Phrase Reinforcement) algorithm was particularly designed to summarise microblogs,
however it is merely able to generate a single-post summary that conveys a single topic, potentially overlooking other important
information from the posts. In this paper, we contribute the PRICE (Phrase Reinforcement: Iteration, Clustering and Extraction)
algorithm by extending the original PR algorithm with the ability to generate both a multi-post summary and a single-post
summary that span over multiple subtopics. Experimental evaluation results show that the PRICE algorithm outperforms the
original PR algorithm in terms of both the ROUGE-1 and the Content metrics.
Paper#29: Camilo Cruz, Justyna Karakiewicz and Michael Kirley.
Generation and exploration of architectural form using a composite Cellular Automata
25
Abstract: In this paper, we introduce a composite Cellular Automata (CA) to explore digital morphogenesis in architecture.
Consisting of multiple interleaved one dimensional CA, our model evolves the boundaries of spatial units in cross sectional
diagrams. We investigate the efficacy of this approach by systematically varying initial conditions and transition rules. Simulation
experiments show that the composite CA can generate aggregate spatial units to match the characteristics of specific spatial
configurations, using a well-known architectural landmark as a benchmark. Significantly, spatial patterns emerge as consequence
of the evolution of the system, rather than from prescriptive design decisions.
Paper#33: Yuyu Liang, Mengjie Zhang and Will Browne.
Wrapper Feature Construction for Figure-ground Image Segmentation Using Genetic Programming
Abstract: Abstract. Figure-ground segmentation is a process of separating regions of interest from unimportant backgrounds. It is
challenging to separate objects from target images with high variations (e.g. cluttered backgrounds), which requires effective
feature sets to capture the discriminative information between object and background regions. Feature construction is a process of
transforming a given set of features to a new set of high-level features, which considers the interactions between the previous
features, thus the constructed features can be more meaningful and effective. As Genetic programming (GP) is a well-suited
algorithm for feature construction (FC), it is employed to conduct both multiple FC (MFC) and single FC (SFC), which aims to
improve the segmentation performance for the first time in this paper. The cooperative coevolution technique is introduced in GP
to construct multiple features from different types of image features separately while conducting feature combination
simultaneously, called as CoevoGPMFC. One wrapper method (wrapperGPSFC) is also designed, and one well-performing
embedded method (embeddedGPSFC) is introduced as a reference method. Compared with the original features extracted by
existing feature descriptors, the constructed features from the proposed methods are more robust and performance better on the
test set. Moreover, the features constructed by the three methods achieve similar performance for the given segmentation tasks.
Paper#34: Francesco Fico, Francesco Urbino, Robert Carrese, Pier Marzocca and Xiaodong Li.
Surrogate-assisted Multi-swarm Particle Swarm Optimization of Morphing Airfoils
Abstract: This paper presents a study to design, analyze and optimize an airfoil trailing edge, i.e., shape morphing of the airfoil
trailing-edge topology. The primary idea behind morphing is to improve the wing performance for different flight conditions.
Modern aircrafts are designed for unique operating conditions. In order to obtain the best configuration, a dynamic optimization
algorithm has been developed based on a Multi-Swarm Particle Swarm Optimization algorithm (MPSO), a population-based
stochastic optimization algorithm inspired by the social interaction among insects or animals. However, with respect to aircraft
design and in the context of computational fluid dynamics (CFD), function evaluations are computationally expensive; typically
requiring large computational grids to obtain a reasonable representation of the flow-field. In this paper, the developed MPSO
algorithm is combined with a Kriging surrogate representation of the objective space, to alleviate the computational effort. The
topology of the trailing edge is defined and characterized by four control points. Two different hypothetical mission profiles are
analyzed. The results exhibit an improvement of around 2% with respect to the original airfoil for every flight condition treated.
Paper#35: Xuhui Zhang, Kevin Korb, Ann Nicholson and Steven Mascaro.
Applying Dependency Patterns in Causal Discovery of Latent Variable Models
Abstract: Latent variables represent unmeasured causal factors. Some, such as intelligence, cannot be directly measured; others
may be, but we do not know about them or know how to measure them when making our observations. Regardless, in many cases,
the influence of latent variables is real and important, and optimal modeling cannot be done without them. However, in many of
those cases the influence of latent variables reveals itself in patterns of measured dependency that cannot be reproduced using the
observed variables alone, under the assumptions of the causal Markov property and faithfulness. In such cases, latent variables
may be posited to the advantage of the causal discovery process. All latent variable discovery takes advantage of this; we make
the process explicit.
Paper#37: Mohammad Haqqani, Xiaodong Li and Xinghuo Yu.
An Evolutionary Multi-criteria Journey Planning Algorithm for Multimodal Transportation Networks
Abstract: Journey planners are considered as one of the promising solutions for enhancing the transportation quality in urban
cities, hence reducing the congestion and environmental pollution. In this context, however, passengers are not just mere users of
the systems, instead represents an active component having their own preferences towards traveling that needs to be satisfied.
Thus, in this work, we propose a novel evolutionary-based journey planner that incorporates users' preferences into the journey
planner. The experimental results demonstrated that the proposed approach, is able to provide a good quality journeys that
are relevant to the travelers.
Paper#38: Mohammad Haqqani, Xiaodong Li and Xinghuo Yu.
Estimating Passenger Preferences Using Implicit Relevance Feedback for Personalized Journey
Abstract: Nowadays, personalized journey planning is becoming increasingly popular, due to a strong practical interest and
higher-quality route suggestion in align with commuter preferences. In a journey planning system, individuals are not just mere
users of the systems, instead represents an active component willing to take different routes based on their own preferences, e.g.,
26
the fastest, least changes or cheapest journey. In this work, we propose a novel preference estimation method that incorporates
implicit relevance feedback methods into journey planner. Our method utilizes commuters travel history to generate the
corresponding preference model. The model is adaptive and will be updated iteratively during the user/planner interactions. Using
experiments with real dataset, we demonstrate that our suggested method can improve journeys quality even in absence of explicit
rating from users.
Paper#41: Muhammad Durrani and Jong-Myon Kim.
Quantitative Assessment of Heart’s Function: A Hybrid Mechanism for Left Ventricle’s Segmentation from Cine MRI
Sequences
Abstract: In this paper, we propose a hybrid approach for segmenting the left ventricle out of magnetic resonance sequences and
results of the segmentation are used for heart quantification. In the hybrid approach, a thresholding based region growing
algorithm coupled with gradient vector flow (GVF) is used for the desired task. Results of the segmentation steps are used for
quantification process and yielded 175.4 ± 51.52, 66 ± 38.97 and 61.60 ± 12.79 for End Diastolic Volume (EDV), End Systolic
Volume (ESV) and Ejection Fraction (EF), respectively.
Paper#42: Manjurul Islam, Rashedul Islam and Jong-Myon Kim.
A Hybrid feature selection scheme based on local compactness and global separability for improving roller bearing
diagnostic performance
Abstract: This paper proposes a hybrid feature selection scheme for identifying the most discriminant fault signatures using an
improved class separability criteria—the local compactness and global separability (LCGS)—of distribution in feature dimension
to diagnose bearing faults. The hybrid model consists of filter based selection and wrapper based selection. In the filter phase, a
sequential forward floating selection (SFFS) algorithm is employed to yield a series of suboptimal feature subsets candidates
using LCGS based feature subset evaluation metric. In the wrapper phase, the most discriminant feature subset is then selected
from suboptimal feature subsets based on maximum average classification accuracy estimation of support vector machine (SVM)
classifier using them. Effectiveness of the proposed hybrid feature selection method is verified with fault diagnosis application for
low speed rolling element bearings under various conditions. Experimental results indicate that the proposed method outperforms
the state-of-the-art algorithm, yielding 1.4 % to 17.74% diagnostic performance improvement in average classification accuracy.
Paper#43: Dileep Appana, Rashedul Islam and Jong-Myon Kim.
Reliable Fault Diagnosis of Bearings Using Distance and Density Similarity on an Enhanced k-NN
Abstract: The k-nearest neighbor (k-NN) is a simple and highly effective classifier. However, in multi-class classification
problems, where the density of data samples varies across different classes, the classification accuracy of k-NN degrades and
becomes highly sensitive to the neighborhood size, k. This is mainly due to its use of distance based measure of similarity
between different samples. We propose a density-weighted distance similarity metric to improve the classification accuracy of
standard k-NN, which considers both distance between the samples, and their relative densities. The performance of the proposed
k-NN is not affected by the neighborhood size, k. We use the proposed k-NN classifier to improve the diagnostic performance of a
fault diagnosis scheme for rolling element bearings. Experimental results show that it yields better classification accuracy as
compared to traditional k-NN.
Paper#44: Bo Yuan and Tiantian Zhang.
Towards Solving TSPN with Arbitrary Neighborhoods: A Hybrid Solution
Abstract: As the generalization of TSP (Travelling Salesman Problem), TSPN (TSP with Neighborhoods) is closely related to
several important real-world applications. However, TSPN is significantly more challenging than TSP as it is inherently a mixed
optimization task containing both combinatorial and continuous components. Different from previous studies where TSPN is
either tackled by approximation algorithms or formulated as a mixed integer problem, we present a hybrid framework in which
metaheuristics and classical TSP solvers are combined strategically to produce high quality solutions for TSPN with arbitrary
neighborhoods. The most distinctive feature of our solution is that it imposes no explicit restriction on the shape and size of
neighborhoods, while many existing TSPN solutions require the neighborhoods to be disks or ellipses. Furthermore, various
continuous optimization algorithms and TSP solvers can be conveniently adopted as necessary. Experiment results show that,
using two off-the-shelf routines and without any specific performance tuning efforts, our method can efficiently solve TSPN
instances with up to 25 regions, which are represented by both convex and concave random polygons.
Paper#45: Abdelmonaem Abdallah, Daryl Essam and Ruhul Sarker.
Detectable Genetic Algorithms-based techniques for solving Dynamic Optimisation Problem with Unknown Active
Variables
Abstract: A dynamic Optimisation Problem with Unknown Active Variables (DOPUAV) is a dynamic problem, in which the
activeness of the variables changes as time passes, to simulate the dynamicity in the problem’s variables. In this paper, several
variations of genetic algorithms are proposed to solve DOPUAV. They are called Detectable techniques. These techniques try to
detect where the problem changes, before detecting the active variables. These variations are tested, then the best variation is
27
compared with the best previously used algorithms namely Hyper Mutation (HyperM), Random Immigration GA (RIGA), as well
as simple GA (SGA). The results and statistical analysis show the superiority of our proposed algorithm.
Paper#46: Boxiong Tan, Hai Huang, Hui Ma and Mengjie Zhang.
Binary PSO for Web Service Location-Allocation
Abstract: Web services are independently programmable application components which scatter over the Internet. Network
latency is one of the major concerns of web service application. Thus, physical locations of web services and users should be
taken into account for web service composition. In this paper, we propose a new solution based on the modified binary PSO-based
(MBPSO) approach which employs an adaptive inertia technique to allocating web service locations. Although several heuristic
approaches have been proposed for web service location-allocation, to our best knowledge, this is the first time applying PSO to
solve the problem. A simulated experiment is done using the WS-DREAM dataset with five different complexities. To compare
with genetic algorithm and original binary PSO approaches, the proposed MBPSO approach has advantages in most situations.
Paper#47: Ayad Turky, Nasser R. Sabar and Andy Song.
Neighbourhood analysis: a case study on Google Machine Reassignment Problem
Abstract: It is known that neighbourhood structures affect search performance. In this study we analyse a series of
neighbourhood structures to facilitate the search. The well known steepest descent (SD) local search algorithm is used in this
study as it is parameter free. The search problem used is the Google Machine Reassignment Problem (GMRP). GMRP is a recent
real world problem proposed at ROADEF/EURO challenge 2012 competition. It consists in reassigning a set of services into a set
of machines for which the aim is to improve the machine usage while satisfying numerous constraints. In this paper, the
effectiveness of three neighbourhood structures and their combinations are evaluated on GMRP instances, which are very diverse
in terms of number of processes, resources and machines. The results show that neighbourhood structure does have impact on
search performance. A combined neighbourhood structures with SD can achieve results better than SD with single neighbourhood
structure.
Paper#48: Anupam Trivedi, Dipti Srinivasan, Kunal Pal and Thomas Reindl.
A MOEA/D with Non-uniform Weight Vector Distribution Strategy for Solving the Unit Commitment Problem in
Uncertain Environment
Abstract: In this paper, a multiobjective evolutionary algorithm based on decomposition (MOEA/D) based is proposed to solve
the unit commitment (UC) problem in uncertain environment as a multi-objective optimization problem considering cost,
emission, and reliability as the multiple objectives. The uncertainties occurring due to thermal generator outage and load forecast
error are incorporated using expected energy not served (EENS) reliability index and EENS cost is used to reflect the reliability
objective. Since, UC is a mixed-integer optimization problem, a hybrid strategy is integrated within the framework of
decomposition-based MOEA such that genetic algorithm (GA) evolves the binary variables while differential evolution (DE)
evolves the continuous variables. To enhance the performance of the presented algorithm, novel non-uniform weight vector
distribution strategies are proposed. The effectiveness of the non-uniform weight vector distribution strategy is verified through
stringent simulated results on different test systems.
