Welcome Message from General Chairsdmbd2020.ic-si.org/program/program-icsi-dmbd2020.pdfthe ICSI-DMBD’2020 will provide you with such excellent program and schedule. The technical

ICSI 2020 & DMBD 2020, July 14 - 15, 2020, Belgrade, Serbia

Welcome Message from General Chairs

We are warmly welcoming you from all over the world to attend the 11th International

Conference on Swarm Intelligence (ICSI’2020) held in conjunction with the 5th International

Conference on Data Mining and Big Data (DMBD’2020) in Belgrade of Serbia.

The theme of the ICSI-DMBD’2020 is “Serving Life with Swarm Intelligence and Data

Science”. On the one hand, ICSI 2020 was the 11th international gathering in the world

for researchers working on most of aspects of swarm intelligence, following successful events in

Chiang Mai (ICSI 2019), Shanghai (ICSI 2018), Fukuoka (ICSI 2017), Bali (ICSI 2016), Beijing

(ICSI-CCI 2015), Hefei (ICSI 2014), Harbin (ICSI 2013), Shenzhen (ICSI 2012), Chongqing

(ICSI 2011), and Beijing (ICSI 2010), which provided a high-level academic forum for

participants to disseminate their new research findings and discuss emerging areas of research.

It also created a stimulating environment for participants to interact and exchange information

on future challenges and opportunities in the field of swarm intelligence research. On the

other hand, the DMBD 2020 is the 5th event after the successful first event (DMBD’2016)

at Bali Island of Indonesia, second event (DMBD’2017) at Fukuoka City of Japan, third

event (DMBD’2018) at Shanghai of China, and fourth event (DMBD’2019) at Chiang Mai

of Thailand. With the advent of big data analysis and intelligent computing techniques we

are facing new challenges to make the information transparent and understandable efficiently.

The DMBD 2020 provided an excellent opportunity and an academic forum for academia

and practitioners to present and discuss the latest scientific results, methods, and innovative

ideas and advantages in theories, technologies, and applications in data mining, big data, and

intelligent computing.

The ICSI-DMBD’2020 will provide a good opportunity for academia and practitioners

to present and discuss the latest scientific results and methods, the innovative ideas and

advantages in theories, technologies and applications in both swarm intelligence and data

mining for big data.

The ICSI’2020 technical program will cover all aspects of swarm intelligence and related

areas, while the DMBD’2020 technical program will cover many aspects of data mining and

big data. The co-location events definitely benefit both fields of swarm intelligence and data

mining for big data and will stimulate some innovative ideas in the cutting-edges of those areas.

We believe that you will enjoy this important and hard-to-get gathering for the communities

of the swarm intelligence and data mining for big data.

Thanks to the hard work of the Organization Committees and the Program Committees,

the ICSI-DMBD’2020 will provide you with such excellent program and schedule. The technical

1

program will cover many important aspects in swarm intelligence and data mining for big data.

Due to the pandemic of COVID-19, even though the ICSI-DMBD 2020 was planned

to be held at Singindunum University in Belgrade of Serbia, and after carefully evaluating

most of announcements and guidance regarding to COVID-19, and too much restrictions on

oversea-traveling, released by relevant national departments, the ICSI-DMBD 2020 organizing

committee has made the decision that our ICSI-DMBD 2020 will continue as scheduled on

July 14-19, by being converted to a fully virtual conference. The ICSI-DMBD 2020 technical

team will be providing the ability for the authors with accepted papers to present their work

through an interactive online platform or video replay. The presentations of accepted authors

will be made available to all registered attendees online.

In addition, the ICSI-DMBD’2020 will definitely contribute a lot to the enhancement of the

research horizons of our delegates in both fields of swarm intelligence and data science.

On behalf of the organizing and technical committees, I wish the ICSI-DMBD’2019 will be

a memorable event for you.

Sincerely yours!

General Chairs of ICSI-DMBD’2020

Ying Tan

Peking University, China

Milan Tuba

Singidunum University, Serbia

2


Welcome Message from Program Committee Chair

The 11th International Conference on Swarm Intelligence (ICSI’2020) is the 11th international

gathering in the world for researchers working on all aspects of swarm intelligence, following the

successful and fruitful previous ten events (ICSI’2019-2016, ICSI-CCI’2015, ICSI’2014-2010),

which provided an excellent opportunity and/or an academic forum for academics and

practitioners to present and discuss the latest scientific results and methods, innovative ideas,

and advantages in theories, technologies, and applications in swarm intelligence. In this year

event, the ICSI’2020 will be held in conjunction with the 5th International Conference on Data

Mining and Big Data (DMBD’2020) at Belgrade of Serbia for sharing common mutual ideas,

promoting transverse fusion, and stimulating innovation. The aim of this important co-location

events are to exhibit the state of the art research and development in many aspects of both

swarm intelligence and data mining from theoretical to practical researches.

For ICSI’2020 event, it received 127 submissions and invited submissions from about 291

authors in 24 countries and regions (Brazil, Bulgaria, Cameroon, Canada, China, Colombia,

Ecuador, Germany, Greece, India, Iran, Italy, Japan, Mexico, Peru, Russia, Serbia, Slovakia,

Taiwan of China, Thailand, Turkey, United Kingdom, United States, Venezuela) across six

continents (Asia, Europe, North America, South America, Africa, and Oceania). Each

submission was reviewed by at least two reviewers, and on average 2.4 reviewers. Based on

rigorous reviews by the Program Committee members and reviewers, 63 high-quality papers

were selected for publication in this proceedings volume with an acceptance rate of 49.6%.

The papers are organized in 12 cohesive sections covering major topics of swarm intelligence

research and its development and applications.

For DMBD’2020 event, it received received 39 submissions and invited manuscripts from

about 91 authors in 11 countries and regions (Brunei Darussalam, China, Colombia, Cuba,

India, Japan, Malaysia, Russia, South Korea, Thailand, Venezuela). Each submission was

reviewed by at least two reviewers, and on average 3.1 reviewers. Based on rigorous reviews

by the Program Committee members and reviewers, 10 high-quality papers were selected for

publication in this proceedings volume with an acceptance rate of 25.64%. The contents of

those papers are covering some major topics of data mining and big data.

On behalf of the Organizing Committee of ICSI-DMBD 2020, we would like to

express sincere thanks to International Association of Swarm and Evolutionary Intelligence

(IASEI)(iasei.org), which is the premier international scholarly society devoted to advancing

the theories, algorithms, real-world applications and developments of swarm intelligence

and evolutionary intelligence, for its sponsorship, to Peking University, Southern University

3

of Science and Technology and Singindunum University for their co-sponsorship, and to

Computational Intelligence Laboratory of Peking University and IEEE Beijing Chapter for its

technical co-sponsorship, as well as to our supporters of International Neural Network Society,

World Federation on Soft Computing, Beijing Xinghui Hi-Tech Co. and Springer-Nature.

We would also like to thank the members of the Advisory Committee for their guidance,

the members of the international Program Committee and ad- ditional reviewers for reviewing

the papers, and the members of the Publications Committee for checking the accepted papers

in a short period of time. We are particularly grateful to the proceedings publisher Springer

for publishing the proceedings in the prestigious series of Lecture Notes in Computer Science.

Moreover, we wish to express our heartfelt appreciation to the plenary speakers, session chairs,

and student helpers. In addition, there are still many more colleagues, associates, friends, and

supporters who helped us in immeasurable ways; we express our sincere gratitude to them

all. Last but not the least, we would like to thank all the speakers, authors, and participants

for their great contributions that made ICSI-DMBD 2020 successful and all the hard work

worthwhile.

We sincerely hope that all ICSI-DMBD’2020 participants will enjoy attending conference

sessions and social activities, meeting research partners, and setting up new research

collaborations.

Cheers!

ICSI-DMBD’2020 General Program Committee Chair

Yuhui Shi

Southern University of Science and Technology, China

4


Contents

Messages 1

Welcome Message from General Chairs . . . . . . . . . . . . . . . . . . 1

Welcome Message from Program Committee Chair . . . . . . . . . . . 3

Organizing Committees 6

Organizing Committees . . . . . . . . . . . . . . . . . . . . . . . . . . 6

International Program Committee . . . . . . . . . . . . . . . . . . . . . 8

Sponsors 12

Program Schedule and Technical Program Overview 13

Technical Program 16

ICSI-DMBD 2020 Oral Sessions, July 14 (Tue) . . . . . . . . . . . . . . 16

ICSI-DMBD 2020 Oral Sessions, July 15 (Wed) . . . . . . . . . . . . . 23

Abstracts 26

Index 48

5

Organizing Committees

General Co-chairs

Ying Tan, Peking University, China,

Milan Tuba, Singidunum University, Serbia

Programme Committee Chair

Yuhui Shi, Southern University of Science and Technology, China

Advisory Committee Co-chairs

Milovan Stanisic, Singidunum University, Serbia

Gary G. Yen, Oklahoma State University, USA

Russell C. Eberhart, Oklahoma State University, USA

Technical Committee Co-chairs

Haibo He, University of Rhode Island Kingston, USA

Kay Chen Tan, City University of Hong Kong, China

Nikola Kasabov, Aukland University of Technology, New Zealand

Ponnuthurai Nagaratnam Suganthan, Nanyang Technological University, Singapore

Xiaodong Li, RMIT University, Australia

Hideyuki Takagi, Kyushu University, Japan

Mengjie Zhang, Victoria University of Wellington, New Zealand

M.Middendorf, University of Leipzig, Germany

Mengjie Zhang, Victoria University of Wellington, New Zealand

Plenary Session Co-chairs

A.Engelbrecht, University of Pretoria, South Africa

Chaoming Luo, University of Detroit Mercy, USA

Invited Session Co-chairs

Andres Iglesias, University of Cantabria, Spain

Haibin Duan, Beihang University, China

Junfeng Chen, Hohai University, China

Special Session Co-chairs

Ben Niu, Shenzhen University, China

Yan Pei, University of Aizu, Japan

Qirong Tang, Tongji University, China

Tutorial Co-chairs

Shi Cheng, Shanxi Normal University, China

Junqi Zhang, Tongji University, China

Yinan Guo, China University of Mining and Technology, China

6


Publication Co-chairs

Swagatam Das, Indian Statistical Institute, India

Radu-Emil Precup, Politehnica University of Timisoara, Romania

Finance and Registration Co-chairs

Andreas Janecek, University of Vienna, Austria

Suicheng Gu, Google Corporation, USA

Publicity Co-Chairs Yew-Soon Ong, Nanyang Technological University, Singapore

Carlos Coello, CINVESTAV-IPN, Mexico

Yaochu Jin, University of Surrey, UK

Rossi Kamal, GERIOT, Bangladesh

Dongbin Zhao, Institute of Automation, CAS, China

Local Arrangement Chair

Mladen Veinovic, Singidunum University, Serbia

Nebojsa Bacanin, Singidunum University, Serbia

Eva Tuba, Singidunum University, Serbia

7

International Program Committee

ICSI International Program Committee

Ashik (Islamic University of Technology) Ahmed (Islamic University of Technology)

Abdelmalek (Tahar Moulay University of Saida) Amine (Tahar Moulay University of Saida)

Esther (INTA) Andres (INTA)

Sabri (Istanbul University) Arik (Istanbul University)

Nebojsa (Singidunum university) Bacanin (Singidunum university)

Carmelo J. A. (University of Pernambuco) Bastos Filho (University of Pernambuco)

Sandeep (G.S. nstitute of Technology) Bhongade (G.S. nstitute of Technology)

Sujin (Khon Kaen University) Bureerat (Khon Kaen University)

David (Universidad Politecnica de Madrid) Camacho (Universidad Politecnica de Madrid)

Bin (Tsinghua University) Cao (Tsinghua University)

Abdelghani (Universite des Sciences et Technologie

d’Oran)

Chahmi (Universite des Sciences et Technologie d’Oran)

Walter (National Taipei University of Technology) Chen (National Taipei University of Technology)

Mu-Song (Da-Yeh University) Chen (Da-Yeh University)

Long (Institute of Automation, Chinese Academy of

Science)

Cheng (Institute of Automation, Chinese Academy of

Science)

Prithviraj (U. S. Naval Research Laboratory) Dasgupta (U. S. Naval Research Laboratory)

Haibin (Beijing University of Aeronautics and

Astronautics)

Duan (Beijing University of Aeronautics and

Astronautics)

Andries (University of Stellenbosch) Engelbrecht (University of Stellenbosch)

Amir H. (University of Technology, Sydney) Gandomi (University of Technology, Sydney)

Hongyuan (Harbin Engineering University) Gao (Harbin Engineering University)

Shangce (University of Toyama) Gao (University of Toyama)

Ping (Beijing Normal University) Guo (Beijing Normal University)

Ahmed (University of Djelfa) Hafaifa (University of Djelfa)

Guosheng (Jiangsu Normal University) Hao (Jiangsu Normal University)

Weiwei (Peking University) Hu (Peking University)

Changan (Ritsumeikan University) Jiang (Ritsumeikan University)

Mingyan (Shandong University) Jiang (Shandong University)

Colin (University of Nottingham) Johnson (University of Nottingham)

Yasushi (Nippon Institute of Technology) Kambayashi (Nippon Institute of Technology)

Vivek (Universita degli Studi di Cagliari) Kumar (Universita degli Studi di Cagliari)

Xiujuan (Shaanxi Normal University) Lei (Shaanxi Normal University)

Bin (University of Science and Technology of China) Li (University of Science and Technology of China)

Jing (Zhengzhou University) Liang (Zhengzhou University)

Ju (Shandong University) Liu (Shandong University)

Wenlian (Fudan University) Lu (Fudan University)

Chaomin (Mississippi State University) Luo (Mississippi State University)

Wenjian (University of Science and Technology of

China)

Luo (University of Science and Technology of China)

Chengying (Jiangxi University of Finance and

Economics)

Mao (Jiangxi University of Finance and Economics)

Sreeja (PSG College of Technology) N.K (PSG College of Technology)

Endre (Singidunum University) Pap (Singidunum University)

Yan (University of Aizu) Pei (University of Aizu)

Thomas (ORNL) Potok (ORNL)

Radu-Emil (Politehnica University of Timisoara) Precup (Politehnica University of Timisoara)

Boyang (Zhong Yuan University) Qu (Zhong Yuan University)

Ivana (Singidunum university) Strumberger (Singidunum university)

Ponnuthurai (Nanyang Technological University) Suganthan (Nanyang Technological University)

8


Ying (Peking University) Tan (Peking University)

Akash (IGDTUW) Tayal (IGDTUW)

Eva (University of Belgrade) Tuba (University of Belgrade)

Mladen (Singidunum University) Veinovic (Singidunum University)

Guoyin (Chongqing University of Posts and

Telecommunications)

Wang (Chongqing University of Posts and

Telecommunications)

Yan (The Ohio State University) Wang (The Ohio State University)

Benlian (Changshu Institute of Technology) Xu (Changshu Institute of Technology)

Yu (Nanjing University of Information Science and

Technology)

Xue (Nanjing University of Information Science and

Technology)

Yingjie (De Montfort University) Yang (De Montfort University)

Peng-Yeng (National Chi Nan University) Yin (National Chi Nan University)

Jun (Kyushu University) Yu (Kyushu University)

Ling (Jinan University) Yu (Jinan University)

Jie (Newcastle University) Zhang (Newcastle University)

Qieshi (Shenzhen Institutes of Advanced Technology,

Chinese Academy of Sciences)

Zhang (Shenzhen Institutes of Advanced Technology,


Junqi (Tongji University) Zhang (Tongji University)

Xinchao (Beijing University of Posts and

Telecommunications)

Zhao (Beijing University of Posts and

Telecommunications)

Miodrag (Singidunum University) Zivkovic (Singidunum University)

Dejan (Singidunum University) Zivkovic (Singidunum University)

9

DMBD International Program Committee

Miltos (University of Texas at San Antonio) Alamaniotis (University of Texas at San Antonio)

Nebojsa (Singidunum university) Bacanin (Singidunum university)

Carmelo J. A. (University of Pernambuco) Bastos Filho (University of Pernambuco)

Tossapon (Mae Fah Luang University) Boongoen (Mae Fah Luang University)

David (Universidad Politecnica de Madrid) Camacho (Universidad Politecnica de Madrid)

Abdelghani (Universite des Sciences et Technologie

d’Oran)

Chahmi (Universite des Sciences et Technologie d’Oran)

Vinod (University of Kerala) Chandra S. S. (University of Kerala)

Hui (Liverpool John Moores University) Cheng (Liverpool John Moores University)

Jose Alfredo Ferreira (Federal University, UFRN) Costa (Federal University, UFRN)

Bei (Shaanxi Nomal University) Dong (Shaanxi Nomal University)

Qinqin (Shanghai Maritime University) Fan (Shanghai Maritime University)

A.H. (Stevens Institute of Technology) Gandomi (Stevens Institute of Technology)

Liang (Huazhong Univ. of Sci. & Tech.) Gao (Huazhong Univ. of Sci.& Tech.)

Shangce (University of Toyama) Gao (University of Toyama)

Teresa (Universidad Estatal da Peninsula de Santa

Elena - UPSE)

Guarda (Universidad Estatal da Peninsula de Santa

Elena - UPSE)

Weian (Tongji University) Guo (Tongji University)

Weiwei (Peking University) Hu (Peking University)

Dariusz (Wroc?aw University of Technology) Jankowski (Wroc?aw University of Technology)

Mingyan (Shandong University) Jiang (Shandong University)

Qiaoyong (Xi’an University of Technology) Jiang (Xi’an University of Technology)

Chen (Hohai University) Junfeng (Hohai University)

Imed (LCOMS - Universite de Lorraine) Kacem (LCOMS - Universite de Lorraine)

Kalinka (University of Sofia) Kaloyanova (University of Sofia)

Vivek (Universita degli Studi di Cagliari) Kumar (Universita degli Studi di Cagliari)

Bin (University of Science and Technology of China) Li (University of Science and Technology of China)

Qunfeng (Dongguan University of Technology) Liu (Dongguan University of Technology)

Wenjian (University of Science and Technology of

China)

Luo (University of Science and Technology of China)

Wojciech (Wroclaw University of Technology) Macyna (Wroclaw University of Technology)

Katherine (University of South Africa) Malan (University of South Africa)

Vasanth Kumar (SCSVMV University) Mehta (SCSVMV University)

Yi (Victoria University of Wellington) Mei (Victoria University of Wellington)

Efren (University of Veracruz) Mezura-Montes (University of Veracruz)

Sheak Rashed Haider (Daffodil International

University)

Noori (Daffodil International University)

Endre (Singidunum University) Pap (Singidunum University)

Mario (University of Catania) Pavone (University of Catania)

Yan (University of Aizu) Pei (University of Aizu)

Somnuk (Universiti Teknologi Brunei) Phon-Amnuaisuk (Universiti Teknologi Brunei)

Manik (indian) Sharma (indian)

Pramod Kumar (ABV-IIITM Gwalior) Singh (ABV-IIITM Gwalior)

Joao (GECAD) Soares (GECAD)

Ivana (Singidunum university) Strumberger (Singidunum university)

Yifei (Shaanxi Normal University) Sun (Shaanxi Normal University)

Hung-Min (National Tsing Hua University) Sun (National Tsing Hua University)

Ying (Peking University) Tan (Peking University)

Paulo (ISEL) Trigo (ISEL)

Milan (Singidunum University) Tuba (Singidunum University)

Eva (University of Belgrade) Tuba (University of Belgrade)

Agnieszka (Warsaw University of Technology) Turek (Warsaw University of Technology)

10


Mladen (Singidunum University) Veinovic (Singidunum University)

Gai-Ge (China Ocean University) Wang (China Ocean University)

Guoyin (Chongqing University of Posts and

Telecommunications)

Wang (Chongqing University of Posts and

Telecommunications)

Zhenzhen (Jinling Institute of Technology) Wang (Jinling Institute of Technology)

Yan (The Ohio State University) Wang (The Ohio State University)

Ka-Chun (City University of Hong Kong) Wong (City University of Hong Kong)

Rui (Hohai University) Xu (Hohai University)

Zhile (Shenzhen Institute of Advanced Technology,


Yang (Shenzhen Institute of Advanced Technology,


Yingjie (De Montfort University) Yang (De Montfort University)

Wei-Chang (Department of Industrial Engineering and

Engineering Management)

Yeh (Department of Industrial Engineering and

Engineering Management)

Peng-Yeng (National Chi Nan University) Yin (National Chi Nan University)

Jie (Newcastle University) Zhang (Newcastle University)

Qieshi (Shenzhen Institutes of Advanced Technology,


Zhang (Shenzhen Institutes of Advanced Technology,


Junqi (Tongji University) Zhang (Tongji University)

Xinchao (Beijing University of Posts and

Telecommunications)

Zhao (Beijing University of Posts and

Telecommunications)

11

Sponsors

Sponsors

International Association of Swarm and

Evolutionary Intelligence

Co-Sponsors

Peking University

Southern University of Science and Technology

Singidunmum University

Computational Intelligence Laboratory, Peking

University

Technical Sponsors

IEEE Computational Intelligence Society

Technical Co-Sponsors

International Neural Network

Society

World Federation of Soft

Computing

SpringerLecture Notes in Computer

Science

Beijing Xinghui High-Tech Co.

12


Program Schedule and Technical Program Overview

Notable Events

Date Time Event

July 14 (Tue) 09:00 - 12:00 Parallel Oral Sessions

(zoom: Room I and Room II)

12:00 - 13:00 Break

13:00 - 21:00 Parallel Oral Sessions

(Zoom: Room I and Room II)

July 15 (Wed) 10:00 - 11:20 Parallel Oral Sessions

(zoom: Room I and Room II)

11:20 - 14:00 Break

13:00 - 15:40 Parallel Oral Sessions

(Zoom: Room I and Room II)

15:40 - 18:00 Award Ceremony

(Zoom: Room I)

Technical Program Overview

13

Room I Room II

July

14

(Tue)

09:00 - 10:20

Genetic Algorithm and Evolutionary

Computation

Data Mining

Machine Learning

Other Applications

10:20 - 10:40 Break

10:40 - 12:00 Bacterial Foraging Optimization

Bacterial Foraging Optimization

Brain Storm Optimization Algorithm

12:00 - 13:00 Break

13:00 - 14:20

Ant Colony Optimization

Multi-agent System and Robotic Swarm


Computation

Machine Learning

Data Mining

Multi-Objective Optimization

Other Applications

14:20 - 14:40 Break

14:40 - 16:00

Ant Colony Optimization

Swarm Intelligence and Nature-Inspired

Computing

Particle Swarm Optimization

16:00 - 16:20 Break

16:20 - 17:40 Other Applications


Computation

Other Applications

17:40 - 18:00 Break

18:00 - 19:20

Machine Learning


Particle Swarm Optimization


Computing

Other Applications

19:20 - 19:40 Break

19:40 - 21:00


Computing

Swarm-based Computing Algorithms for

Optimization

Brain Storm Optimization Algorithm


Computation

Other Applications

July

15

(Wed)

10:00 - 11:20 Swarm-based Computing Algorithms for

Optimization

Swarm-based Computing Algorithms for

Optimization

Other Applications

11:20 - 14:00 Break

14:00 - 15:40

Data Mining

Machine Learning

Multi-agent System and Robotic Swarm


Technical Program

Oral Sessions

July 14, 2020(Tuesday)

Date July 14, 2020(Tuesday) Location Room I Time 09:00-10:20

09:00 - 09:20 A New Local Search Adaptive Genetic Algorithm for the

Pseudo-Coloring Problem

P26

Rodrigo Contreras, Orides Morandin Junior and Monique Viana

09:20 - 09:40 Method based on Data Mining Techniques for Breast Cancer

Recurrence Analysis

P26

Morales-Ortega Roberto Cesar, Lozano-Bernal German,

Ariza-Colpas Paola Patricia, Arrieta-Rodriguez Eugenia,

Ospino-Mendoza Elisa Clementina, Caicedo-Ortiz Jose,

Pineres-Melo Marlon Alberto, Mendoza-Palechor Fabio Enrique and

Roca-Vides Margarita

09:40 - 10:00 Aula Touch Game: Digital tablets and their incidence in the

development of citizen competences of middle education students

in the district of Barranquilla-Colombia.

P26

Paola Patricia Ariza Colpas, Belina Annery Herrera Tapias, Andres

Gabriel Sanchez Comas, Marlon Alberto Pineres Melo and Judith

Martinez Royert

10:00 - 10:20 Parasite-Guest Infection Modeling: Social Science Applications P27

Cesar Vargas-Garcıa, Jenny Paola Lis-Gutierrez, Mercedes

Gaitan-Angulo and Melissa Lis-Gutierrez

Date July 14, 2020(Tuesday) Location Room II Time 09:00-10:20

09:00 - 09:20 Site Selection of the Colombian Antarctic Research Station based on

Fuzzy-Topsis Algortihm

P27

Jairo Coronado Hernandez, wilson rios angulo, camilo segovia, diana

urrego nino and alfonso romero

16


09:20 - 09:40 Use of the industrial property system in Colombia (2018): a machine

learning application

P27

Jenny Paola Lis Gutierrez, melissa lis gutierrez, adriana patricia

gallego torres and manuel romero

09:40 - 10:00 Econometric Algorithms Applied to the Incidence of Income on

Satisfaction with Quality of Life in Latin American Capitals

P27

Carolina Henao, mercedes gaitan, Jenny Paola Lis-Gutierrez and

leonor mojica

10:00 - 10:20 Hypothesis Verification for Designing Flyback Booster by Analysis

of Variance Visualized on Triangular Matrix Representations

P28

Kazuhisa Chiba, Taiki Hatta and Masahiro Kanazaki


10:40 - 11:00 Bacterial Foraging Optimization Based on Levy Flight for Fuzzy

Portfolio Optimization

P28

Xinzheng Wu, Tianwei Zhou and Zishan Qiu

11:00 - 11:20 Modified Bacterial Foraging Optimization for Fuzzy

Mean-Semivariance-Skewness Portfolio Selection

P29

Xinzheng Wu, Aiqing Gao and Xin Huang

11:20 - 11:40 An Improved Bacterial Foraging Optimization with Dierential

and Poisson Distribution Strategy and its Application to Nurse

Scheduling Problem

P29

Jingzhou Jiang, Xiaojun Xiong, Yikun Ou and Hong Wang

11:40 - 12:00 Improved Bacterial Foraging Optimization Algorithm with

Comprehensive Swarm Learning Strategies

P29

Xiaobing Gan and Baoyu Xiao


10:40 - 11:00 Adaptive bacterial foraging optimization based on roulette strategy P28

Weifu Cao, Yingsi Tan, Miaojia Huang and Yuxi Luo

11:00 - 11:20 An Adapting Chemotaxis Bacterial Foraging Optimization

Algorithm for Feature Selection in Classification

P29

Hong Wang and Yikun Ou

17

11:20 - 11:40 BSO-CLS: Brain Storm Optimization Algorithm with Cooperative

Learning Strategy

P30

Liang Qu, Qiqi Duan, Jian Yang, Shi Cheng, Ruiqi Zheng and Yuhui

Shi

11:40 - 12:00 A Hybrid Brain Storm Optimization Algorithm for Dynamic Vehicle

Routing Problem

P30

Mingde Liu, Yang Shen and Yuhui Shi


13:00 - 13:20 An Ant Colony Optimization Algorithm Based Automated

Generation of Software Test Cases

P30

Saju Sankar S and Vinod Chandra S S

13:20 - 13:40 A Multi-agent Ant Colony Optimization Algorithm for Effective

Vehicular Traffic Management

P31


13:40 - 14:00 A Structural testing model using SDA algorithm P31


14:00 - 14:20 Computational Analysis of Third-Grade Liquid Flow with Cross

Diffusion Effects: Application to Entropy Modeling

P31

K Loganathan, A. Charles Sagayaraj, amelec viloria, noel varela,

omar bonerge and luis ortiz ospino


13:00 - 13:20 Deep Learning Strategies for Survival Prediction in Prophylactic

Resection Patients

P32

Anand Hareendran S, Vinodchandra S.S., Sreedevi Prasad and

Dhanya S

13:20 - 13:40 Optimal Reservoir Optimization Using Multiobjective Genetic

Algorithm

P32

Anand Hareendran S., Vinodchandra S.S. and Saju Sankar S.

13:40 - 14:00 Analytical Study of Radiative Casson Nanoliquid Flow with Heat

Absorption

P32

K Loganathan, k Tamilvanan, amelec viloria, noel varela and omar

bonerge

18


14:00 - 14:20 Multi-objective Particle Swarm Optimisation for Cargo Packaging in

Large Containers

P32

Anand Hareendran S., Vinodchandra S.S. and Saju Sankar S.


14:40 - 15:00 Hybrid Ant Colony Optimization-based Method for Focal of a

Disease Segmentation in Lung CT Images

P33

Mingli Lu

15:00 - 15:20 An Ant-Inspired Track-to-Track Recovery Approach for

Construction of Cell Lineage Trees

P33

Di Wu, Hui Bu, Benlian Xu, Mingli Lu and Zhen Sun

15:20 - 15:40 Swarm Intelligence in Data Science: Applications, Opportunities and

Challenges

P33

Jian Yang, Liang Qu, Yang Shen, Yuhui Shi, Shi Cheng, Junfeng

Zhao and Xiaolong Shen

15:40 - 16:00 A Two-Step Approach to the Search of Minimum Energy Designs

via Swarm Intelligence

P34

Frederick Kin Hing Phoa and Tzu-Chieh Tsai


14:40 - 15:00 Special Session 1- A Performance Class-Based Learning Particle

Swarm Optimization

P34

Chia Emmanuel Tungom, Maja Gulan and Ben Niu

15:00 - 15:20 Optimizing Hydrography Ontology Alignment through Compact

Particle Swarm Optimization Algorithm

P34

Yifeng Wang, Hanguang Yao, Liangpeng Wan, Hua Li, Junjun

Jiang, Yun Zhang, Fangmin Wu, Junfeng Chen, Xingsi Xue and

Cai Dai

15:20 - 15:40 Research on Crowd-sensing Task Assignment Based on Fuzzy

Inference PSO Algorithm

P35

Jianjun Li, Jia Fu, Yu Yang, Xiaoling Wang and Xin Rong

19

15:40 - 16:00 The improvement of V-shaped transfer function of Binary Particle

Swarm Optimization

P35

Dong-Yang Zhang, Jian-Hua Liu, Lei Jiang, Guan-Nan Bu,

Ren-Yuan Hu and Yi-Xuan Luo


16:20 - 16:40 PCA Based Kernel Initialization for Convolutional Neural Networks P35

Yifeng Wang, Junfeng Chen, Yuxi Rong, Hongyue Pan, Ke Liu,

Yang Hu, Fangmin Wu, Wei Peng and Xingsi Xue

16:40 - 17:00 Research on Short-term Urban Traffic Congestion Based on Fuzzy

Comprehensive Evaluation and Machine Learning

P36

Yuan Mei, Ting Hu and Lichun Yang

17:00 - 17:20 Adaptive and Dynamic Knowledge Transfer in Multi-task Learning

with Attention Networks

P36

Tao Ma and Ying Tan

17:20 - 17:40 Application of decision tree algorithm based on clustering and

entropy method level division for regional economic index selection

P36

Yi Zhang and Gang Yang


16:20 - 16:40 An Improved CMA-ES for Solving Large Scale Optimization

Problem

P37

Jin Jin, Chuan Yang and Yi Zhang

16:40 - 17:00 A Genetic Algorithm-Based Solver for small-scale Jigsaw Puzzles P37

Wenjing Guo, Wenhong Wei, Yuhui Zhang and Anbing Fu

17:00 - 17:20 A New EDA with Dimension Reduction Technique for Large Scale

Many-objective Optimization

P37

Mingli Shi, Lianbo Ma and Guangming Yang

17:20 - 17:40 Newtonian heating effects of Oldroyd-B liquid flow with

cross-difussion and second order slip

P37

K Loganathan, k Tamilvanan, amelec viloria, noel varela and omar

bonerge

20



18:00 - 18:20 Imbalanced Ensemble Learning for Enhanced Pulsar Identification P38

Jakub Holewik, Gerald Schaefer and Iakov Korovin

18:20 - 18:40 Image Clustering by Generative Adversarial Optimization and

Advanced Clustering Criteria

P38

Eva Tuba, Ivana Strumberger, Nebojsa Bacanin, Timea Bezdan and

Milan Tuba

18:40 - 19:00 Methods of Machine Learning in System Abnormal Behavior

Detection

P38

Pavel Savenkov and Alexey Ivutin

19:00 - 19:20 Success-history based parameter adaptation in MOEA/D algorithm P38

Shakhnaz Akhmedova and Vladimir Stanovov


18:00 - 18:20 RPP algorithm: a method for discovering interesting rare itemsets P39

Sadeq Darrab, David Broneske and Gunter Saake

18:20 - 18:40 Pareto optimization in oil refinery P39

Dmitri Kostenko, Dmitriy Arseniev, Vyacheslav Shkodyrev and

Vadim Onufriev

18:40 - 19:00 Map Generation and Balance in the Terra Mystica Board Game

Using Particle Swarm and Local Search

P39

Luiz Jonata Pires De Araujo, Alexandr Grichshenko, Rodrigo

Lankaites Pinheiro, Rommel D. Saraiva and Susanna Gimaeva

19:00 - 19:20 Prediction of Photovoltaic Power using Nature-Inspired Computing P39

Miroslav Sumega, Anna Bou Ezzeddine, Gabriela Grmanova and

Viera Rozinajova


19:40 - 20:00 On Assessing the Temporal Characteristics of Reaching the

Milestone by a Swarm

P40

Eugene Larkin and Maxim Antonov

20:00 - 20:20 Synchronized Swarm Operation P40

Eugene Larkin, Tatyana Akimenko and Aleksandr Privalov

21

20:20 - 20:40 Colour Quantisation by Human Mental Search P40

Seyed Jalaleddin Mousavirad, Gerald Schaefer, Hui Fang, Xiyao Liu

and Iakov Korovin

20:40 - 21:00 A Novel Image Segmentation Based on Clustering and

Population-Based Optimisation

P40

Seyed Jalaleddin Mousavirad, Gerald Schaefer, Hossein

Ebrahimpour-Komleh and Iakov Korovin


19:40 - 20:00 Determinative Brain Storm Optimization P41

Georgia Sovatzidi and Dimitris Iakovidis

20:00 - 20:20 Archive Update Strategy Influences Differential Evolution

Performance

P41

Vladimir Stanovov, Shakhnaz Akhmedova and Eugene Semenkin

20:20 - 20:40 Analysis of Breast Cancer detection using different Machine learning

techniques

P41

Siham A. Mohammed, Sadeq Darrab, Salah A. Noaman and Gunter

Saake

22


July 15, 2020(Wednsday)

Date July 15, 2020(Wednsday) Location Room I Time 10:00-11:20

10:00 - 10:20 Learning Automata-based Fireworks Algorithm On Adaptive

Assigning Sparks

P42

Zhang Junqi, Che Lei and Chen Jianqing

10:20 - 10:40 A Novel Biogeography-based Optimization Algorithm with

Momentum Migration and Taxonomic Mutation

P42

Xinchao Zhao, Yisheng Ji and Junling Hao

10:40 - 11:00 Binary Pigeon-Inspired Optimization for Quadrotor Swarm

Formation Control

P42

Zhiqiang Zheng, Haibin Duan and Chen Wei

11:00 - 11:20 The Research of Flexible Scheduling of Workshop Based on Artificial

Fish Swarm Algorithm and Knowledge Mining

P42

Jieyang Peng, Jiahai Wang, Dongkun Wang, Andreas Kimmig and

Jivka Ovtcharova

Date July 15, 2020(Wednsday) Location Room II Time 10:00-11:20

10:00 - 10:20 A Modified Artificial Bee Colony Algorithm for Scheduling

Optimization of Multi-Aisle AS/RS System

P43

Xiaohui Yan, Felix T. S. Chan, Zhicong Zhang, Cixing Lv and Shuai

Li

10:20 - 10:40 Canine Algorithm for Node Disjoint Paths P43

R Ananthalakshmi Ammal, P C Sajimon and Vinod Chandra S. S.

10:40 - 11:00 Research on PM2.5 Integrated Prediction Model Based on

lasso-RF-GAM

P43

Yan Peng, Tingxian Wu, Ziru Zhao and Haoxiang Wei

11:00 - 11:20 An Evaluation Algorithm of the Importance of Network Node Based

on Community Influence

P44

Gongzhen He, Junyong Luo and Meijuan Yin

23

Date July 15, 2020(Wednsday) Location Room I Time 14:00-15:40

14:00 - 14:20 Target Tracking Algorithm based on Density Clustering P44

Chen Jin Long, Zeng Qinghao and Qin Xingguo

14:20 - 14:40 A Method for Localization and Classification of Breast Ultrasound

Tumors

P44

Wanying Mo, Yuntao Zhu and Chaoyun Wang

14:40 - 15:00 Case Classification Processing and Analysis Method for Respiratory

Belt Data

P45

Jinlong Chen and Mengke Jiang

15:00 - 15:20 A Tool for Supporting the Evaluation of Active Learning Activities P45

Waraporn Jirapanthong

15:20 - 15:40 Inferring Candidate CircRNA-disease Associations by Bi-random

Walk Based on CircRNA Regulatory Similarity

P45

Chunyan Fan, Xiujuan Lei and Ying Tan

Date July 15, 2020(Wednsday) Location Room II Time 14:00-15:40

14:00 - 14:20 O-flocking: Optimized Flocking Model on Autonomous Navigation

for Robotic Swarm

P46

Li Ma, Weidong Bao, Xiaomin Zhu, Meng Wu, Yuan Wang,

Yunxiang Ling and Wen Zhou

14:20 - 14:40 A Parallel Evolutionary Algorithm with Value Decomposition for

Multi-Agent Problems

P46

Gao Li, Qiqi Duan and Yuhui Shi

14:40 - 15:00 Research on Sliding Mode Control of Underwater

Vehicle-manipulator System Based on an Exponential Approach

Law

P46

Qirong Tang, Yang Hong, Zhenqiang Deng, Daopeng Jin and

Yinghao Li

15:00 - 15:20 Multi-objective Combinatorial Generative Adversarial Optimization

and Its Application in Crowdsensing

P46

Yi-nan Guo, Jianjiao Ji, Ying Tan and Shi Cheng

24


15:20 - 15:40 Multi-Objective Dynamic Scheduling Model of Flexible Job Shop

Based on NSGAII Algorithm and Scroll Window Technology

P47

Yingli Li and Jiahai Wang

25

Abstracts

July 14, 2020(Tuesday) Room I 09:00-10:20

A New Local Search Adaptive Genetic Algorithm for thePseudo-Coloring Problem

Rodrigo Contreras, Orides Morandin Junior and Monique VianaAbstract. Several applications result in a gray level image partitioned into different regions ofinterest. However, the human brain has difficulty in recognizing many levels of gray. In some cases,this problem is alleviated with the attribution of artificial colors to these regions, thus configuring anapplication in the area of visualization and graphic processing responsible for categorizing samplesusing colors. However, the task of making a set of distinct colors for these regions stand out isa problem of the NP-hard class, known as the pseudo-coloring problem (PsCP). In this work,it is proposed to use the well-known meta-heuristic Genetic Algorithm together with operatorsspecialized in the local search for solutions as well as self-adjusting operators responsible for guidingthe parameterization of the technique during the resolution of PsCPs. The proposed methodologywas evaluated in two different scenarios of color assignment, having obtained the best results incomparison to the techniques that configure the state of the art.

Method based on Data Mining Techniques for Breast CancerRecurrence Analysis

Morales-Ortega Roberto Cesar, Lozano-Bernal German, Ariza-Colpas Paola Patricia,Arrieta-Rodriguez Eugenia, Ospino-Mendoza Elisa Clementina, Caicedo-Ortiz Jose,

Pineres-Melo Marlon Alberto, Mendoza-Palechor Fabio Enrique and Roca-Vides MargaritaAbstract. Cancer is a constantly evolving disease, which affects a large number of people worldwide.Great efforts have been made at the research level for the development of tools based on data miningtechniques that allow to detect or prevent breast cancer. The large volumes of data play a fundamentalrole according to the literature consulted, a great variety of dataset oriented to the analysis of thedisease has been generated, in this research the Breast Cancer dataset was used, the purpose of theproposed research is to submit comparison of the J48 and randomforest, NaiveBayes and NaiveBayesSimple, SMO Poli-kernel and SMO RBF-Kernel classification algorithms, integrated with the SimpleK-Means cluster algorithm for the generation of a model that allows the successful classification ofpatients who are or Non-recurring breast cancer after having previously undergone surgery for thetreatment of said disease, finally the methods that obtained the best levels were SMO Poly-Kernel+ Simple K-Means 98.5% of Precision, 98.5 % recall, 98.5% TPRATE and 0.2% FPRATE. Theresults obtained suggest the possibility of using intelligent computational tools based on data miningmethods for the detection of breast cancer recurrence in patients who had previously undergone surgery.

Aula Touch Game: Digital tablets and their incidence in thedevelopment of citizen competences of middle education students

in the district of Barranquilla-Colombia.Paola Patricia Ariza Colpas, Belina Annery Herrera Tapias, Andres Gabriel Sanchez Comas,

Marlon Alberto Pineres Melo and Judith Martinez RoyertAbstract. Citizen competences are considered as a fundamental aspect in the social developmentof man with his environment, which allows him to carry out actions that are articulated with thedifferent guidelines established by law, which leads the citizen to live in a coherent and peacefulway in a nation that tends for freedom of thought framed in a democratic society. That is why itis considered of high importance that in educational establishments there are spaces that tend fortraining in peaceful coexistence framed in the law of the educated. This article resulted from theresearch project: “Social Appropriation of citizen and mathematical competences making use ofMIDTablets”, in which the mediation of Information and Communication Technologies is proposed tosupport the training of citizens with competences citizens who ensure adequate behavior in society.This project was developed in 31 educational institutions in the district of Barranquilla-Colombia,with support from resources of both the Ministry of Information Technology and Communications

26


(MinTic), and the Secretariat of District Education of Barranquilla, in compliance with national goals, departmental and district regarding the quality of education of the national population

Parasite-Guest Infection Modeling: Social Science ApplicationsCesar Vargas-Garcıa, Jenny Paola Lis-Gutierrez, Mercedes Gaitan-Angulo and Melissa

Lis-GutierrezAbstract. In this study we argue that parasite-host infections are a major research topic because oftheir implications for human health, agriculture and wildlife. The evolution of infection mechanisms isa research topic in areas such as virology and ecology. Mathematical modelling has been an essentialtool to obtain a better systematic and quantitative understanding of the processes of parasiticinfection that are difficult to discern through strictly experimental approaches. In this article wereview recent attempts using mathematical models to discriminate and quantify these infectionmechanisms. We also emphasize the challenges that these models could bring to new fields of studysuch as social sciences and economics.

July 14, 2020(Tuesday) Room II 09:00-10:20

Site Selection of the Colombian Antarctic Research Station basedon Fuzzy-Topsis Algortihm

Jairo Coronado Hernandez, wilson rios angulo, camilo segovia, diana urrego nino and alfonsoromero

Abstract. By 2025 the Republic of Colombia aims to be an advisory member of the AntarcticTreaty System (ATS) and the installation of a scientific station is necessary to upscale the scientificcapabilities. The aim of this paper is showing the results of the implementation of a Fuzzy TOPSISalgorithm for site selection of the Colombian Antarctic Scientific Station. A three-phase methodologywas proposed, and the obtained results allowed to identify the optimum location for the station,considering key success factors and regulatory constraints.

Use of the industrial property system in Colombia (2018): amachine learning application

Jenny Paola Lis Gutierrez, melissa lis gutierrez, adriana patricia gallego torres and manuelromero

Abstract. The purpose of this paper is to establish ways to predict the spatial distribution of theuse of the intellectual property system from information on industrial property applications andgrants (distinctive signs and new creations) and copyright registrations in 2018. This will be doneusing supervised learning algorithms applied to information on industrial property applications andgrants (trademarks and new creations) and copyright registrations in 2018. Within the findings, 4algorithms were identified with a level of explanation higher than 80%: (i) Linear Regression, withan elastic network regularization; (ii) Stochastic Gradient Descent, with Hinge loss function, Ringeregularization (L2) and a constant learning rate; (iii) Neural Networks, with 1,000 layers, with Adam’ssolution algorithm and 2,000 iterations; (iv) Random Forest, with 10 trees.

Econometric Algorithms Applied to the Incidence of Income onSatisfaction with Quality of Life in Latin American Capitals

Carolina Henao, mercedes gaitan, Jenny Paola Lis-Gutierrez and leonor mojicaAbstract. The purpose of this research is to determine the incidence of income on the satisfaction ofquality of life in Latin American capitals. The data was taken from the CAF Survey in 2017 (ECAF),made by Development Bank of Latin America. To do this, an econometric model was constructedwhich represents the relationship between the studied variables. Among the main findings, it wasidentified that, except for Mexico City, in all the analyzed cities a deteriorated state of health reducesthe probability that the individual feels satisfied with his or her life. Therefore, health is a crucialelement to increase the citizens’ perception of the quality of life in Latin American cities.

27

Hypothesis Verification for Designing Flyback Booster byAnalysis of Variance Visualized on Triangular Matrix

RepresentationsKazuhisa Chiba, Taiki Hatta and Masahiro Kanazaki

Abstract. This study performed data mining for nondominated-solution datasets of flyback-boostergeometry for next-generation space transportation procured by evolutionary computation. Weprepared two datasets of nondominated solutions due to two problem definitions, which differmerely in the definition of some design variables based on a design hypothesis gained fromevolutionary-computation results. This study aims at verifying the hypothesis by applying miningto these two datasets to elucidate the contrast in the influence of the design variables. We usedfunctional analysis of variance for data mining; scrutinized the effects of single and two-combineddesign variables. Furthermore, intuitive visualization by triangular matrix representations coulddistinguish the discrepancy between the obtained results. The consequence has verified the significanceof the hypothesis; it revealed that the discontinuous surface naturally evaded in the hypersonic rangebecause of surface temperature upsurge is capable of enhancing the lift-to-drag ratio in the low-speedrange; the hypothesis grew into a new design problem.


Bacterial Foraging Optimization Based on Levy Flight for FuzzyPortfolio Optimization

Xinzheng Wu, Tianwei Zhou and Zishan QiuAbstract. In this paper, a new kind of bacterial foraging optimization that combines withlevy flight (LBFO) is employed to solve a novel portfolio optimization (PO) problem with fuzzyvariables and modified mean-semivariance model which includes the transaction fee (includingthe purchase fee and sell fee), no short sales and the original proportion of the different assets.First of all, a chemotaxis step size using levy distribution takes the place of fixed chemotaxisstep size, which makes a good balance between local search and global search through frequentshort-distance search and occasional long-distance search. Moreover, fuzzy variables are used tosignify the uncertainty of future risks and returns on assets and some constrained conditions aretaken into consideration. The results of the simulation show that the model can be solved morereasonably and effectively by LBFO algorithm than the original bacterial foraging optimization (BFO).

Adaptive bacterial foraging optimization based on roulettestrategy

Weifu Cao, Yingsi Tan, Miaojia Huang and Yuxi LuoAbstract. Bacterial foraging optimization has drawn great attention and has been applied widelyin various fields. However, BFO performs poorly in convergence when coping with more complexoptimization problems, especially multimodal and high dimensional tasks. Aiming to address theseissues, we therefore seek to propose a hybrid strategy to improve the BFO algorithm in each stageof the bacteria’s’ foraging behavior. Firstly, a non-linear descending strategy of step size is adoptedin the process of flipping, where a larger step size is given to the particle at the very beginning ofthe iteration, promoting the rapid convergence of the algorithm while later on a smaller step size isgiven, helping enhance the particles’ global search ability. Secondly, an adaptive adjustment strategyof particle aggregation is introduced when calculating step size of the bacteria’s swimming behavior.In this way, the particles will adjust the step size according to the degree of crowding to achieveefficient swimming. Thirdly, a roulette strategy is applied to enable the excellent particles to enjoyhigher replication probability in the replication step. A linear descent elimination strategy is adoptedfinally in the elimination process. The experimental results demonstrate that the improved algorithmperforms well in both single-peak function and multi-peak function, having strong convergence abilityand search ability.

28


An Improved Bacterial Foraging Optimization with Dierentialand Poisson Distribution Strategy and its Application to Nurse

Scheduling ProblemJingzhou Jiang, Xiaojun Xiong, Yikun Ou and Hong Wang

Abstract. Bacterial Foraging Optimization(BFO) has been predominately applied to some real-worldproblems, but this method has poor convergence speed over complex optimization problems. Inthis paper, an improved Bacterial Foraging Optimization with Differential and Poisson Distributionstrategies(PDBFO) is proposed to promote the insufficiency of BFO. In PDBFO, the step size ofbacteria is segmented and adjusted in accordance with fitness value to accelerate convergence andenhance the search capability. Moreover, the differential operator and the Poisson Distributionstrategy are incorporated to enrich individual diversity, which prevents algorithm from being trappedin the local optimum. Experimental simulations on eleven benchmark functions demonstrate thatthe proposed PDBFO has better convergence behavior in comparison to other six algorithms.Additionally, to verify the effectiveness of the method in solving the real-world complex problems, thePDBFO is also applied to the Nurse Scheduling Problem(NSP). Results indicate that the proposedPDBFO is more effective in obtaining the optimal solutions by comparing with other algorithms.

Improved Bacterial Foraging Optimization Algorithm withComprehensive Swarm Learning Strategies

Xiaobing Gan and Baoyu XiaoAbstract. Bacterial foraging optimization (BFO), a novel bio-inspired heuristic optimizationalgorithm, has been attracted widespread attention and widely applied to various practicaloptimization problems. However, the standard BFO algorithm exists some potential deficiencies,such as the weakness of convergence accuracy and a lack of swarm communication. Owing tothe improvement of these issues, an improved BFO algorithm with comprehensive swarm learningstrategies (LPCBFO) is proposed. As for the LPCBFO algorithm, each bacterium keeps on movingwith stochastic run lengths based on linear-decreasing Levy flight strategy. Moreover, illuminatedby the social learning mechanism of PSO and CSO algorithm, the paper incorporates cooperativecommunication with the current global best individual and competitive learning into the original BFOalgorithm. To examine the optimization capability of the proposed algorithm, six benchmark functionswith 30 dimensions are chosen. Finally, experimental results demonstrate that the performance of theLPCBFO algorithm is superior to the other five algorithms.


Modified Bacterial Foraging Optimization for FuzzyMean-Semivariance-Skewness Portfolio Selection

Xinzheng Wu, Aiqing Gao and Xin HuangAbstract. In this paper, a novel bacterial foraging optimization with decreasing chemotaxis stepcombined with sine function is employed to solve a fuzzy portfolio optimization with a modifiedmean-semivariance-skewness model which includes the transaction fee and no short sales. Firstof all, a decreasing chemotaxis step combined with sine function (BFO-SDC) takes the placeof constant chemotaxis step size. It is a nonlinear decreasing strategy at every iteration of thealgorithm. And then, the variance is replaced by semivariance and skewness is taken into accountin order to generate asymmetry of return distributions to overcome the inadequacy of the standardmean-variance model. Finally, fuzzy variables are used to express the uncertain and impreciseelements in the decision-making process. The results of the simulation show that the model can besolved more reasonably and effectively by BFO-SDC than the original bacterial foraging optimization.

An Adapting Chemotaxis Bacterial Foraging OptimizationAlgorithm for Feature Selection in Classification

Hong Wang and Yikun OuAbstract. Efficient classification methods can improve the data quality or relevance to better

29

optimize some Internet applications such as fast searching engine and accurate identification.However, in the big data era, difficulties and volumes of data processing increase drastically. Todecrease the huge computational cost, heuristic algorithms have been used. In this paper, an AdaptingChemotaxis Bacterial Foraging Optimization (ACBFO) algorithm is proposed based on basic BacterialForaging Optimization (BFO) algorithm. The aim of this work is to design a modified algorithmwhich is more suitable for data classification. The proposed algorithm has two updating strategiesand one structural changing. First, the adapting chemotaxis step updating strategy is responsible toincrease the flexibility of searching. Second, the feature subsets updating strategy better combinesthe proposed heuristic algorithm with the KNN classifier. Third, the nesting structure of BFO hasbeen simplified to reduce the computation complexity. The ACBFO has been compared with BFO,BFOLIW and BPSO by testing on 12 widely used benchmark datasets. The result shows thatACBFO has a good ability of solving classification problems and gets higher accuracy than the othercomparation algorithm.

BSO-CLS: Brain Storm Optimization Algorithm withCooperative Learning Strategy

Liang Qu, Qiqi Duan, Jian Yang, Shi Cheng, Ruiqi Zheng and Yuhui ShiAbstract. Brain storm optimization algorithms (BSO) have shown great potential in manyglobal black-box optimization problems. However, the existing BSO variants can suffer from threeproblems: (1) large-scale optimization problem; (2) hyperparameter optimization problem; (3) highcomputational cost of the clustering operations. To address these problems, in this paper, we proposea simple yet effective BSO variant named Brain Storm Optimization Algorithm with CooperativeLearning Strategy (BSO-CLS). It is inspired by the new ideas generating process of brain storm inwhich the participators propose their own ideas by cooperatively learning other participators’ ideas.Thus, BSO-CLS iteratively updates the candidate solutions by linearly combining other solutionswith the weights deriving from the fitness values of other solutions. To validate the effectiveness of theproposed method, we test it on 6 benchmark functions with the 1000 dimensions. The experimentalresults show that BSO-CLS can outperform the vanilla BSO and the other BSO variant with thelearning strategy.

A Hybrid Brain Storm Optimization Algorithm for DynamicVehicle Routing Problem

Mingde Liu, Yang Shen and Yuhui ShiAbstract. The Dynamic Vehicle Routing Problem (DVRP) has many real-world applications andpractical values. The objective of DVRP is to find the optimal routes for a fleet of vehicles to servicethe given customer requests, without violating the vehicle capacity constraint. In this paper, a hybridalgorithm is proposed for solving the DVRP with the objective to minimize the total distance ofthe vehicles. The Brain Storm Optimization in objective space (BSO-OS) is applied to guide thechoice of different strategies for the periodic reoptimization of routes. In the BSO-OS procedure,Adaptive Large Neighborhood Search (ALNS) and Ant Colony System (ACS) are used to generatenew solutions. The experiments on the DVRP benchmark and comparative studies are conducted,from which 12 out of 21 new best solutions are obtained by the proposed algorithm, and the othernine solutions are also very competitive. The experimental results show that the proposed algorithmis very effective and competitive.


An Ant Colony Optimization Algorithm Based AutomatedGeneration of Software Test Cases

Saju Sankar S and Vinod Chandra S SAbstract. Software testing is an important process of detecting bugs in the software product therebya quality software product is developed. Verification and Validation (V V) activities are the effectivemethods employed to achieve quality. Static and dynamic testing activities are performed during V

30


V. During static testing, the program code is not executed while in dynamic testing (Black Box andWhite Box), the execution of the program code is performed. Effective test cases are designed byboth these methods. Tables are employed to represent test case documentation. The most beneficialrepresentation - State table based testing, for generating test inputs is explained with the help of stategraphs and state tables. This technique is mainly employed in embedded system software testing,real time applications and web application based software product testing where time constraintsare a major criteria. Automatic generation of test cases will help to reduce time overhead in testingactivities. Our study is to develop optimum test cases by a modified Ant Colony Optimization(ACO) technique in an automated method and it ensures maximum coverage. The prediction modelused in this paper ensures better accuracy of the design of test inputs. A comparison of the similaroptimization techniques was also discussed that is used in automated test case generation. A casestudy of the various states during the execution of a task in an operating system has been presentedto illustrate our approach.

A Multi-agent Ant Colony Optimization Algorithm for EffectiveVehicular Traffic Management

Saju Sankar S and Vinod Chandra S SAbstract. An intelligent agent refers to an autonomous entity directing its activity towardsachieving goals, acting upon an environment using data obtained with the help of a sensorymechanism. Intelligent agent software is a software system that performs tasks independently onbehalf of a user in a networking environment based on user interface and past experiences. By thedesign of an intelligent sensing software program we can regulate the flow of traffic in a transportationinfrastructure network. The problems leading to inefficiencies like loss of time, decrease in safety ofvehicles and pedestrians, massive pollution, high wastage of fuel energy, degradation in the qualityof life can be achieved by the optimized design. Ant Colony Optimization (ACO) has proven to bea very powerful optimization model for combinatorial optimization problems. The algorithm has theobjective of regulating high real time traffic enabling every vehicle in the network with increasedefficiency to minimize factors like time delay and traffic congestion.

A Structural testing model using SDA algorithmSaju Sankar S and Vinod Chandra S S

Abstract. Path testing is the most needed and useful coverage criterion in structural testing.Tracing and obtaining the resultant paths is the main problem in path coverage testing. Evolutionarytechniques are adopted in many software product evaluation methods such as generating and selectionof input test data. The priority of the feasible paths is also to be determined. In this paper,we proposes an optimization algorithm for identifying the effective test data execution paths incontrol flow graph for the program module under test and finding the most efficient test paths usingmodified smell detection agent based optimization algorithm. New innovations are being conductedfor bio-motivated algorithmic techniques from the characteristics of animal behavior. Smell detectionagent based algorithm helps to identify most feasible paths and it uses sequential search to obtain allpaths in a graph. The tester achieves the paths to be tested through a number of smell spot valuesfrom the source node to the target node. We will use control flow graph to produce perfect test pathsand cyclomatic complexity number for obtaining the number of feasible test paths. The best feasiblepaths are prioritized using smell detection agent algorithm such that all the paths are thoroughlytested which ensures structural testing. This algorithm generates paths equal to the cyclomaticcomplexity. It can be illustrated that the proposed approach guarantees full path coverage.

Computational Analysis of Third-Grade Liquid Flow with CrossDiffusion Effects: Application to Entropy Modeling

K Loganathan, A. Charles Sagayaraj, amelec viloria, noel varela, omar bonerge and luis ortizospino

Abstract. The key goal of this current study is to analyze the entropy generation with cross diffusioneffects. The third-grade type non-Newtonian fluid model is used in this study. The current flowproblem is modelled with stretching plate. Modified Fourier heat flux is replaced the classical heatflux. The appropriate transformation is availed to convert the basic boundary layers equations intoODEs and then verified by homotopy algorithm. The consequences of various physical quantities on

31

temperature, velocity, entropy and concentration profile are illustrated graphically.


Deep Learning Strategies for Survival Prediction in ProphylacticResection Patients

Anand Hareendran S, Vinodchandra S.S., Sreedevi Prasad and Dhanya SAbstract. Human race is looking forward to an era where science and technology can wipeoutthe threats laid by lethal diseases. Major statistics shows that about 10 million people die fromvarious forms of cancer annually. Every sixth death in the world is caused by cancer. Treatment tocancer always depend on its type and spread. Treatment includes single or combination of surgery,chemotherapy and radiation therapy. In this paper, survival prediction in prophylactic resectionpatients are carried out using various deep learning methods. Prophylactic resection has been foundto be very effective in colon cancer, breast cancer and ovarian cancer. In this paper, we try to validatethe results in a test environment using multi layered deep neural network. Classical Navie Bayer’salgorithm has been used to classify the dataset and convolution neural network (CNN) has been usedto create the survival prediction model. Results affirm better survival results in prophylactic resectionpatients.

Optimal Reservoir Optimization Using Multiobjective GeneticAlgorithm

Anand Hareendran S., Vinodchandra S.S. and Saju Sankar S.Abstract. Scarcity of fresh water resources has thrown various challenges to hydrologist. Optimumusage of resource is the only way out to handle this situation. Among the various water resources themost controllable one is the dam reservoirs. This paper deals with optimal reservoir optimization usingmulti objective genetic algorithm (MOGA). Various parameters like reservoir storage capacity, spillloss, evaporation rate, water used for irrigation, water used for electricity production, rate of inflow,outflow all need to be managed in an optimal way so that water levels are managed and resourcespecifications are met. This is normally managed using a software, but sudden change in scenariosand change in requirements cannot be handled by such softwares. Hence we are incorporating anoptimised software layer to handle such situation. Multi objective genetic algorithm was able tooptimise the water usage within the usage constrains. The results were assessed based on reliability,vulnerability and resilience indices. In addition, based on a multi-criteria decision-making model,it was evaluated by comparing it with other evolutionary algorithms. The simulated result showsthat MOGA derived rules are promising and competitive and can be effectively used for reservoiroptimization operations.

Analytical Study of Radiative Casson Nanoliquid Flow with HeatAbsorption

K Loganathan, k Tamilvanan, amelec viloria, noel varela and omar bonergeAbstract. The divergence of thermally radiative MHD flow of a Casson nanofluid over a stretchingpaper alongside heat absorption. The governing non linear equations are remodeled into a nonlinearODE’s. The HAM is adopted to find the series solution. The changes of pertinent parameters areanalyzed with diagrams and tables. The fluid velocity is controlled by suction and it develops withinjection. The local Nusselt number rapidly suppresses with increasing the magnetic field parameterin heat generation case.

Multi-objective Particle Swarm Optimisation for CargoPackaging in Large Containers

Anand Hareendran S., Vinodchandra S.S. and Saju Sankar S.Abstract. Cargo management in all mode of transports like airlines, ships and trucks is a challengingtask. The way in which an optimal allocation of packages in different containers are done using asoftware controlled method. An agent based software module is enabled as a service for the optimum

32


allocation of cargo packages in the container terminals. There are multiple factors that will affectthis allocation - size, shape, weight of the cargo packets and the container. When we design anoptimal allocation module in a software these components need to be addressed along with capacityof the container. Hence, a multi-objective optimization algorithm will improve the performance ofcargo management software. In this paper we suggest a Mixed Species Particle Swarm Optimisation(MSPSO) procedure for optimal allocation of cargo packages in containers of different size andcapacity. The redesigned version of cargo management software performs well with search spaceon normal time complexity. The simulated results gives an improved optimised allocation thannormalised allocation of cargo packets. The improved implementation performed better in terms ofefficient cargo package allocation.


Hybrid Ant Colony Optimization-based Method for Focal of aDisease Segmentation in Lung CT Images

Mingli LuAbstract. The detection of chest CT scan images of the lung play a key role in clinical decisionmaking for some lung disease, such as tumors, pulmonary tuberculosis, solitary pulmonary nodule,lung masses and so on. In this paper, a novel automated CT scan image segmentation algorithm basedon hybrid Ant Colony algorithm and snake algorithm is proposed. Firstly, traditional snake algorithmis used to detect the possible edge points of focal of a disease. Then Ant Colony Optimization(ACO) algorithm is applied to search the possible edge points of focal of a disease .repeatedly.Finally, real edges can be extracted according to the intensity of pheromones. Simulation experimentresults demonstrate that the proposed algorithm is more efficient and effective than the methods wecompared it to.

An Ant-Inspired Track-to-Track Recovery Approach forConstruction of Cell Lineage Trees

Di Wu, Hui Bu, Benlian Xu, Mingli Lu and Zhen SunAbstract. Correct track-to-track association is crucial to the construction of cell lineage trees aswell as the discovery of novel biological phenomenon that occur at rare frequencies. In this paper,an ant colony optimization based heuristic approach is proposed to link potential tracks throughminimizing the cost function that mainly occurs on the fragmented intervals with the constraint ofmaximum inter-frame displacement. Specifically, both cell motion and morphology are emphasizedin the defined cost function, and two decisions are made respectively to recover the mitotic andnon-mitotic cases. Our method has proven to be feasible that can repair the broken tracklets causedby large migration, occlusion and mitosis missing, as well as false positives and missed detections,and can effectively help the construction of reliable cell lineage trees.

Swarm Intelligence in Data Science: Applications, Opportunitiesand Challenges

Jian Yang, Liang Qu, Yang Shen, Yuhui Shi, Shi Cheng, Junfeng Zhao and Xiaolong ShenAbstract. The Swarm Intelligence (SI) algorithms have been proved to be a comprehensive methodto solve complex optimization problems by simulating the emergence behaviors of biological swarms.Nowadays, data science is getting more and more attention, which needs quick management andanalysis of massive data. Most traditional methods can only be applied to continuous and differentiablefunctions. As a set of population-based approaches, it is proven by some recent research works thatthe SI algorithms have great potential for relevant tasks in this field. In order to gather better insightinto the utilization of these methods in data science and to provide a further reference for futureresearches, this paper focuses on the relationship between data science and swarm intelligence. Afterintroducing the mainstream swarm intelligence algorithms and their common characteristics, boththe theoretical and real-world applications in the literature which utilize the swarm intelligence tothe related domains of data analytics are reviewed. Based on the summary of the existing works, this

33

paper also analyzes the opportunities and challenges in this field, which attempts to shed some lighton designing more effective algorithms to solve the problems in data science for real-world applications.

A Two-Step Approach to the Search of Minimum EnergyDesigns via Swarm IntelligenceFrederick Kin Hing Phoa and Tzu-Chieh Tsai

Abstract. Recently, Swarm Intelligence Based (SIB) method, a nature-inspired metaheuristicoptimization method, has been widely used in many problems that their solutions fall in discrete andcontinuous domains. SIB 1.0 is efficient to converge to optimal solution but its particle size is fixedand pre-defined, while SIB 2.0 allows particle size changes during the procedure but it takes longertime to converge. This paper introduces a two-step SIB method that combines the advantages oftwo SIB methods. The first step via SIB 2.0 serves as a preliminary study to determine the optimalparticle size and the second step via SIB 1.0 serves as a follow-up study to obtain the optimal solution.This method is applied to the search of optimal minimum energy design and the result outperformsthe results from both SIB 1.0 and SIB 2.0.


Special Session 1- A Performance Class-Based Learning ParticleSwarm Optimization

Chia Emmanuel Tungom, Maja Gulan and Ben NiuAbstract. One of the main concerns with Particle Swarm Optimization (PSO) is to increaseor maintain diversity during search in order to avoid premature convergence. In this study, aPerformance Class-Based learning PSO (PCB-PSO) algorithm is proposed, that not only increasesand maintains swarm diversity but also improves exploration and exploitation while speeding upconvergence simultaneously. In the PCB-PSO algorithm, each particle belongs to a class based on itsfitness value and particles might change classes at evolutionary stages or search step based on theirupdated position. The particles are divided into an upper, middle and lower. In the upper class areparticles with top fitness values, the middle are those with average while particles in the bottom classare the worst performing in the swarm. The number of particles in each group is predetermined.Each class has a unique learning strategy designed specifically for a given task. The upper class isdesigned to converge towards the best solution found, Middle class particles exploit the search spacewhile lower class particles explore. The algorithm’s strength is its flexibility and robustness as thepopulation of each class allows us to prioritize a desired swarm behavior. The Algorithm is testedon a set of 8 benchmark functions which have generally proven to be difficult to optimize. Thealgorithm is able to be on par with some cutting edge PSO variants and outperforms other swarmand evolutionary algorithms on a number of functions. On complex multimodal functions, it is ableto outperform other PSO variants showing its ability to escape local optima solutions.

Optimizing Hydrography Ontology Alignment through CompactParticle Swarm Optimization Algorithm

Yifeng Wang, Hanguang Yao, Liangpeng Wan, Hua Li, Junjun Jiang, Yun Zhang, FangminWu, Junfeng Chen, Xingsi Xue and Cai Dai

Abstract. With the explosive growth in generating data in the hydrographical domain, manyhydrography ontologies have been developed and maintained to describe hydrographical featuresand the relationships between them. However, the existing hydrography ontologies are developedwith varying project perspectives and objectives, which inevitably results in the differences in termsof knowledge representation. Determining various relationships between two entities in differentontologies offers the opportunity to link hydrographical data for multiple purposes, though theresearch on this topic is in its infancy. Different from the traditional ontology alignment whosecardinality is 1:1, i.e. one source ontology entity is mapping with one target ontology entity andvice versa, and the relationship is the equivalence, matching hydrography ontologies is a morecomplex task, whose cardinality could be 1:1, 1:n or m:n and the relationships could be equivalence

34


or subsumption. To efficiently optimize the ontology alignment, in this paper, a discrete optimalmodel is first constructed for the ontology matching problem, and then a Compact Particle SwarmOptimization algorithm (CPSO) based matching technique is proposed to efficiently solve it. CPSOutilizes the compact real-value encoding and decoding mechanism and the objective-decomposingstrategy to approximate the PSO’s evolving process, which can dramatically reduce PSO’s memoryconsumption and runtime while at the same time ensure the solution’s quality. The experimentexploits the Hydrography dataset in Complex track provided by the Ontology Alignment EvaluationInitiative (OAEI) to test our proposal’s performance. The experimental results show that CPSO-basedapproach can effectively reduce PSO’s runtime and memory consumption, and determine high-qualityhydrography ontology alignments.

Research on Crowd-sensing Task Assignment Based on FuzzyInference PSO Algorithm

Jianjun Li, Jia Fu, Yu Yang, Xiaoling Wang and Xin RongAbstract. To solve the problem of load unbalance in the case of few users and multi-task, a fuzzyinference PSO algorithm (FPSO) crowd sensing single objective task assignment method is proposed.With task completion time, user load balancing and perceived cost as the optimization goals, thefuzzy learning algorithm dynamically adjusts the learning factor in the PSO algorithm, so that thePSO algorithm can perform global search in the scope of the task space, thus obtaining the optimaltask assignment solution set. Finally, the FPSO algorithm is compared with the PSO, GA and ABCalgorithms on the optimization objectives, such as the algorithm convergence, task completion time,perceived cost and load balance. The experimental results show that the FPSO algorithm not onlyhas faster convergence rate than the other algorithms, and shorten the task completion time, reducethe platform’s perceived cost, improve the user’s load balance, and have a good application effect inthe crowd sensing task assignment.

The improvement of V-shaped transfer function of BinaryParticle Swarm Optimization

Dong-Yang Zhang, Jian-Hua Liu, Lei Jiang, Guan-Nan Bu, Ren-Yuan Hu and Yi-Xuan LuoAbstract. Binary Particle Swarm Optimization (BPSO) is a swarm intelligence to optimizediscrete space problems by extending the Particle Swarm Optimization. Its transfer function isthe key element of BPSO. In this paper, a new V-shaped transfer function with a parameterk has been proposed. The parameter k was used to control the opening size of the transferfunction. At first, the setting of the parameter k has been obtained by the experiments, andthen the new V-shaped transfer with the optimal k value is compared with the other kinds of theV-shaped transfer functions by the experiment of feature selection. The results have indicated thatthe new V-shaped transfer function improved the performance of Binary Particle Swarm Optimization.


PCA Based Kernel Initialization for Convolutional NeuralNetworks

Yifeng Wang, Junfeng Chen, Yuxi Rong, Hongyue Pan, Ke Liu, Yang Hu, Fangmin Wu, WeiPeng and Xingsi Xue

Abstract. The initialization of Convolutional Neural Networks (CNNs) is about providing reasonableinitial values for the convolution kernels and the fully connected layers.. In this paper, we proposeda convolution kernel initialization method based on the two-dimensional principal componentanalysis (2DPCA), in which a parametric equalization normalization method is used to adjust thescale between each neuron weight. After that the weight initial value can be adaptively adjustedaccording to different data samples. This method enables each neuron to fully back-propagateerrors and accelerate network model training. Finally, a network model was built and experimentswere performed using Tanh and ReLU activation functions. The experimental results verify theeffectiveness of the proposed method through the distribution of histograms and the curve comparison

35

diagrams of model training.

Research on Short-term Urban Traffic Congestion Based onFuzzy Comprehensive Evaluation and Machine Learning

Yuan Mei, Ting Hu and Lichun YangAbstract. There are many factors that affect urban traffic flow. In the case of severe trafficcongestion, the vehicle speed is very slow, which results in the GPS positioning system’s estimationof the vehicle speed being very inaccurate, which in turn leads to poor reliability of the estimatedcongestion time of the road segment. The main contents of this study are: in the case of urban trafficcongestion, the prediction and analysis of the degree of traffic congestion and the length of congestion.Taking the dynamic traffic data of Shenzhen on June 9, 2014 as an example, the road section of BinheAvenue is selected, and the data of traffic flow, average speed of traffic volume and traffic volumedensity in the current time period are calculated after data preprocessing, as a measure of traffic Themain impact indicators of congestion status. Then we use the fuzzy comprehensive evaluation methodto divide TSI as a traffic congestion evaluation index and divide the road congestion into four levels.In this way, we can get the congestion of the road in each time period of the day and the time requiredto pass. Then we use the random forest, adaboost, GBDT, Lasso CV and BP neural networks in themachine learning algorithm to build models to measure traffic congestion for training and testing.Finally, the BP neural network has the best effect on this problem, and its model accuracy is 99.22%.Finally, we used BP neural network to predict and congest the road in the next three hours. From theexperimental simulation results, this method can effectively analyze and predict the real-time trafficcongestion.

Adaptive and Dynamic Knowledge Transfer in Multi-taskLearning with Attention Networks

Tao Ma and Ying TanAbstract. Multi-task learning has shown promising results in many applications of machine learning:given several related tasks, it aims to generalize better on the original tasks, by leveraging theknowledge among tasks. The knowledge transfer mainly depends on task relationships. Most ofexisting multi-task learning methods guide learning processes based on predefined task relationships.However, the associated relationships have not been fully exploited in these methods. Replacingpredefined task relationships with the adaptively learned ones may lead to superior performanceas it can avoid the misguiding of improper pre-definition. Therefore, in this paper, we proposeTask Relation Attention Networks to adaptively model the task relationships and dynamicallycontrol the positive and negative knowledge transfer for different samples in multi-task learning. Toevaluate the the effectiveness of the proposed method, experiments on various datasets are conducted.The experimental results demonstrate that the proposed method outperforms both classical andstate-of-the-art multi-task learning baselines.

Application of decision tree algorithm based on clustering andentropy method level division for regional economic index

selectionYi Zhang and Gang Yang

Abstract. The economy of a region is affected by many factors. The purpose of this study is to usethe entropy method clustering and decision tree model fusion to find the main factors affecting theregional economy with the support of big data and empirical evidence. First extract some importantindicators that affect the regional economy, and use the entropy method to find the relative weightsand scores of these indicators. Then use K-means to divide these indicators into several intervals.Based on the entropy fusion model, obtain the ranking of each category of indicators, use theserankings as the objective value of the decision tree, and finally establish an economic indicatorscreening model. Participate in optimization and build a decision tree model that affects regionaleconomic indicators. Through the visualization of the tree and the analysis of feature importance,you can intuitively see the main indicators that affect the regional economy, thereby achieving theresearch goals.

36



An Improved CMA-ES for Solving Large Scale OptimizationProblem

Jin Jin, Chuan Yang and Yi ZhangAbstract. In solving large scale optimization problems, CMA-ES has the disadvantages of highcomplexity and premature stagnation. To solve this problem, this paper proposes an improvedCMA-ES, called GI-ES, for large-scale optimization problems. GI-ES uses all the historicalinformation of the previous generation of individuals to evaluate the parameters of the distribution ofthe next generation. These estimates can be considered as approximate gradient information, whichcomplete covariance information is not required. Thus GI-ES is friendly to large scale optimizationproblems. Comparative experiments have been done on state-of-the-art algorithms. The resultsproved the effectiveness and efficiency of GI-ES for large scale optimization problems.

A Genetic Algorithm-Based Solver for small-scale Jigsaw PuzzlesWenjing Guo, Wenhong Wei, Yuhui Zhang and Anbing Fu

Abstract. In this paper, we present a genetic algorithm-based puzzle solver, which is mainly usedto solve small-scale puzzle problems. We introduce a new measurement function that improves itsaccuracy by normalizing the Mahalanobis distance and the Euclidean distance between two puzzlepieces. By calculating the difference between edges of two puzzle pieces and using the geneticalgorithm to assemble pieces correctly, two ”parent” solutions are merged into one improved ”child”solution. Using the idea of local search, it avoids the problem of local optimum solutions brought bythe genetic algorithm, which greatly improves the accuracy of the puzzle.

A New EDA with Dimension Reduction Technique for LargeScale Many-objective Optimization

Mingli Shi, Lianbo Ma and Guangming YangAbstract. The performance of many-objective evolutionary algorithms deteriorates appreciablyin solving large-scale many-objective optimization problems (MaOPs) which encompass more thanhundreds variables. One of the known rationales is the curse of dimensionality. Estimation ofdistribution algorithms sample new solutions with a probabilistic model built from the statisticsextracting over the existing solutions so as to mitigate the adverse impact of genetic operators. Inthis paper, an Gaussian Bayesian network-based estimation of distribution algorithm (GBNEDA-DR)is proposed to effectively tackle continued large-scale MaOPs. In the proposed algorithm, dimensionreduction technique (i.e. LPP) is employed in the decision space to speed up the estimation searchof the proposed algorithm. The experimental results show that the proposed algorithm performssignificantly better on many of the problems and for different decision space dimensions, and achievescomparable results on some compared with many existing algorithms.

Newtonian heating effects of Oldroyd-B liquid flow withcross-difussion and second order slip

K Loganathan, k Tamilvanan, amelec viloria, noel varela and omar bonergeAbstract. The current study highlights the Newtonian heating and second-order slip velocity withcross-diffusion effectson Oldroyd-B liquid flow. The modified Fourier heat flux is included in theenergy equation system. The present problem is modeled with the physical governing system. Thecomplexity of the governing system was reduced to a nonlinear ordinary system with the help ofsuitable transformations. A homotopy algorithm was used to validate the nonlinear system. Thisalgorithm was solved via MATHEMATICA software. Their substantial aspects are further studiedand reported in detail. We noticed that the influence of slip velocity order two is lower than the slipvelocity order one.

37


Imbalanced Ensemble Learning for Enhanced PulsarIdentification

Jakub Holewik, Gerald Schaefer and Iakov KorovinAbstract. Pulsars can be detected based on their emitted radio waves. Machine learning methodscan be employed to support automated screening of a large number of radio signals for pulsars. Thisis however a challenging task since training these methods is affected by an inherent imbalance in theacquired data with signals relating to actual pulsars being in the minority.

Image Clustering by Generative Adversarial Optimization andAdvanced Clustering Criteria

Eva Tuba, Ivana Strumberger, Nebojsa Bacanin, Timea Bezdan and Milan TubaAbstract. Clustering is the task that has been used in numerous applications including digital imageanalysis and processing. Image clustering refers to the problem of segmenting image for differentpurposes which leads to various clustering criteria. Finding the optimal clusters represented bytheir centers is a hard optimization problem and it is one of the main research focuses on clusteringmethods. In this paper we proposed a novel generative adversarial optimization algorithm for findingthe optimal cluster centers while using standard and advance clustering criteria. The proposedmethod was tested on seven benchmark images and results were compared with the artificial beecolony, particle swarm optimization and genetic algorithm. Based on the obtained results, thegenerative adversarial optimization algorithm founded better cluster centers for image clusteringcompared to named methods from the literature.

Methods of Machine Learning in System Abnormal BehaviorDetection

Pavel Savenkov and Alexey IvutinAbstract. The aim of the research is to develop mathematical and program support for detectingabnormal behavior of users. It will be based on analysis of their behavioral biometric characteristics.One of the major problems in UEBA/DSS intelligent systems is obtaining useful information from alarge amount of unstructured, inconsistent data. Management decision-making should be based onreal data collected from the analysed feature. However, based on the information received, it is ratherdifficult to make any management decision, as the data are heterogeneous and their volumes areextremely large. Application of machine learning methods in implementation of mobile UEBA/DSSsystem is proposed. This will make it possible to achieve a data analysis high quality and find complexdependencies in it. A list of the most significant factors submitted to the input of the analysingmethods was formed during the research.

Success-history based parameter adaptation in MOEA/Dalgorithm

Shakhnaz Akhmedova and Vladimir StanovovAbstract. In this paper two parameter self-adaptation schemes are proposed for the MOEA/D-DEalgorithm. These schemes use the fitness improvement ration to change four parameter values forevery individual separately, as long as in the MOEA/D framework every individual solves its ownscalar optimization problem. The first proposed scheme samples new values and replaces old valueswith new ones if there is an improvement, while the second one keeps a set of memory cells andupdates the parameter values using the weighted sum. The proposed methods are testes on two setsof benchmark problems, namely MOEADDE functions and WFG functions, IGD and HV metrics arecalculated. The results comparison is performed with statistical tests. The comparison shows thatthe proposed parameter adaptation schemes are capable of delivering significant improvements to theperformance of the MOEA/D-DE algorithm. Also, it is shown that parameter tuning is better thanrandom sampling of parameter values. The proposed parameter self-adaptation techniques could beused for other multi-objective algorithms, which use MOEA/D framework.

38



RPP algorithm: a method for discovering interesting rareitemsets

Sadeq Darrab, David Broneske and Gunter SaakeAbstract. The importance of rare itemset mining stems from its ability to discover unseen knowledgefrom datasets in real-life domains, such as identifying network failures, or suspicious behavior. Thereare significant efforts proposed to extract rare itemsets. The RP-growth algorithm outperformsprevious methods proposed for generating rare itemsets. However, the performance of the RP-growthdegrades on sparse datasets, and it is costly in terms of time and memory consumption. Hence,in this paper, we propose the RPP algorithm to extract rare itemsets. The advantage of the RPPalgorithm is that it avoid time for generating useless candidate itemsets by omitting conditionaltrees as RP-growth does. Furthermore, our RPP algorithm uses a novel data structure, RN-list, forcreating rare itemsets. To evaluate the performance of the proposed method, we conduct extensiveexperiments on sparse and dense datasets. The results show that the RPP algorithm is around anorder of magnitude better than the RP-growth algorithm.

Pareto optimization in oil refineryDmitri Kostenko, Dmitriy Arseniev, Vyacheslav Shkodyrev and Vadim Onufriev

Abstract. This article describes the process of multicriteria optimization of a complex industrialcontrol object using Pareto efficiency. The object is being decomposed and viewed as a hierarchy ofembedded orgraphs. Performance indicators and controlling factors lists are created based on theorgraphs and technical specifications of an object, thus allowing to systematize sources of influence.Using statistical data archives to train, the neural network approximates key sensors data to identifythe model of the controllable object and optimize it.

Map Generation and Balance in the Terra Mystica Board GameUsing Particle Swarm and Local Search

Luiz Jonata Pires De Araujo, Alexandr Grichshenko, Rodrigo Lankaites Pinheiro, Rommel D.Saraiva and Susanna Gimaeva

Abstract. Modern board games offer an interesting opportunity for automatically generatingcontent and models for ensuring balance among players. This paper tackles the problem of generatingbalanced maps for a popular and sophisticated board game called Terra Mystica. The complexity ofthe involved requirements coupled with a large search space makes of this a complex combinatorialoptimisation problem which has not been investigated in the literature, to the best of the authors’knowledge. This paper investigates the use of particle swarm optimisation and steepest ascenthill climbing with a random restart for generating maps in accordance with a designed subset ofrequirements. The results of applying these methods are very encouraging, fully showcasing thepotential of search-based metaheuristics in procedural content generation.

Prediction of Photovoltaic Power using Nature-InspiredComputing

Miroslav Sumega, Anna Bou Ezzeddine, Gabriela Grmanova and Viera RozinajovaAbstract. Prediction of photovoltaic (PV) energy is an important task. It allows grid operatorsto plan production of energy in order to secure stability of electrical grid. In this work we focuson improving prediction of PV energy using nature-inspired algorithms for optimization of SupportVector Regression (SVR) models. We propose method, which uses different models optimized forvarious types of weather in order to achieve higher overall accuracy compared to single optimizedmodel. Each sample is classified by Multi-Layer Perceptron (MLP) into some weather class and thenmodel is trained for each weather class. Our method achieved slightly better results compared tosingle optimized model.

39


On Assessing the Temporal Characteristics of Reaching theMilestone by a Swarm

Eugene Larkin and Maxim AntonovAbstract. The physical swarm operation is investigated. Swarm units are considered as three-wheeledmobile robots, moving through rough terrain. For the longitudinal movement of such type vehicle thedynamic model is obtained. The article discusses the issue of forecasting the time a physical swarmreaches a milestone and proposes a hypothesis on the form of the law of the distribution of time duringwhich a swarm unit reaches a milestone. Obtaining this time distribution is carried out with use thePetri-Markov net fundamental apparatus. With use Petri-Markov nets time densities of reaching themilestone both one unit and swarm as a whole are obtained. More common formula of distribution oftime of milestone reaching by l units of K is obtained too. To confirm the hypothesis about the typeof theoretical time distribution a computer experiment was carried out using the Monte Carlo method.

Synchronized Swarm OperationEugene Larkin, Tatyana Akimenko and Aleksandr Privalov

Abstract. Physical swarm system, including number of units, operated in physical time according tocorporative algorithm, is considered. It is shown, that for proper corporative algorithm interpretationit is necessary to synchronize computational processes in units. Structural-parametric model ofsynchronized swarm operation, based on Petri-Markov nets apparatus, is worked out. In thePetri-Markov net transitions are abstract analogues of synchronization procedure, while placessimulate corporative algorithm parts interpretation by swarm units. Primary Petri-Markov modelis transformed into complex semi-Markov process. Formulae for calculation of stochastic and timecharacteristics of the process are obtained. It is shown, that after transformation all methods ofordinary semi-Markov processes investigation may be used for synchronized systems. With use theconcept of distributed forfeit effectiveness of synchronization is evaluated

Colour Quantisation by Human Mental SearchSeyed Jalaleddin Mousavirad, Gerald Schaefer, Hui Fang, Xiyao Liu and Iakov Korovin

Abstract. Colour quantisation is a common image processing technique to reduce the numberof distinct colours in an image which are then represented by a colour palette. The selection ofappropriate entries in this palette is a challenging issue while the quality of the quantised image isdirectly related to the colour palette. In this paper, we propose a novel colour quantisation algorithmbased on the human mental search (HMS) algorithm. HMS is a recent population-based metaheuristicalgorithm with three main operators: mental search to explore the vicinity of candidate solutionsbased on Levy flight, grouping to determine a promising region based on a clustering algorithm,and movement towards the best strategy. The performance of our proposed algorithm is evaluatedon a set of benchmark images and in comparison to four conventional algorithms and seven softcomputing-based colour quantisation algorithms. The obtained experimental results convincinglyshow that our proposed algorithm is capable of outperforming these approaches.

A Novel Image Segmentation Based on Clustering andPopulation-Based Optimisation

Seyed Jalaleddin Mousavirad, Gerald Schaefer, Hossein Ebrahimpour-Komleh and IakovKorovin

Abstract. Image segmentation is an essential step in image processing and computer vision withmany image segmentation algorithms having been proposed in the literature. Among these, clusteringis one of the prominent approaches to achieve segmentation. Traditional clustering algorithms havebeen used extensively for this purpose, although they have disadvantages such as dependence oninitialisation conditions and a tendency to find only local optima. To overcome these disadvantages,population-based metaheuristic algorithms can be applied.

40



Determinative Brain Storm OptimizationGeorgia Sovatzidi and Dimitris Iakovidis

Abstract. Brain Storm Optimization (BSO) is a swarm intelligence optimization algorithm, basedon the human brainstorming process. The ideas of a brainstorming process comprise the solutions ofthe algorithm, which iteratively applies solution grouping, generation and selection operators. Severalmodifications of BSO have been proposed to enhance its performance. In this paper, we propose anovel modification enabling faster convergence of BSO to optimal solutions, without requiring settingan upper bound of algorithm iterations. It considers a brainstorming scenario where participatinggroups with similar ideas recognize that their ideas are similar, and together, collaborate for thedetermination of a better solution. The proposed modification, called Determinative BSO (DBSO),implements this scenario by applying a cluster merging strategy for merging groups of similarsolutions, while following elitist selection. Experimental results using eleven benchmark functionsshow that the proposed modified BSO performs better than both the original and a state-of-the-artalgorithm.

Archive Update Strategy Influences Differential EvolutionPerformance

Vladimir Stanovov, Shakhnaz Akhmedova and Eugene SemenkinAbstract. In this paper the effects of archive set update strategies on differential evolutionalgorithm performance are studied. The archive set is generated from inferior solutions, removedfrom the main population, as the search process proceeds. Next, the archived solutions participatein the search during mutation step, allowing better exploration properties to be achieved. TheLSHADE-RSP algorithm is taken as baseline, and 4 new update rules are proposed, includingreplacing the worst solution, the first found worse solution, the tournament-selected solution andindividually stored solution for every solution in the population. The experiments are performedon CEC 2020 single objective optimization benchmark functions. The results are compared usingstatistical tests. The comparison shows that changing the update strategy significantly improves theperformance of LSHADE-RSP on high-dimensional problems. The deeper analysis of the reasons ofefficiency improvement reveals that new archive update strategies lead to more successful usage ofthe archive set. The proposed algorithms and obtained results open new possibilities of archive usagein differential evolution.

Analysis of Breast Cancer detection using different Machinelearning techniques

Siham A. Mohammed, Sadeq Darrab, Salah A. Noaman and Gunter SaakeAbstract. Data mining algorithms play an important role in the prediction of early-stage breastcancer. In this paper, we propose an approach that improves the accuracy and enhances theperformance of three different classifiers: Decision Tree (J48), Naıve Bayes (NB), and SequentialMinimal Optimization (SMO). We also validate and compare the classifiers on two benchmarkdatasets: Wisconsin Breast Cancer (WBC) and Breast Cancer dataset. Data with imbalanced classesare a big problem in the classification phase since the probability of instances belonging to themajority class is significantly high, the algorithms are much more likely to classify new observationsto the majority class. We address such problem in this work. We use the data level approachwhich consists of resampling the data in order to mitigate the effect caused by class imbalance. Forevaluation, 10 fold cross-validation is performed. The efficiency of each classifier is assessed in termsof accuracy, precision, and recall, true positive and false positive. Experiments show that using aresample filter enhances the classifier’s performance where SMO outperforms others in the WBCdataset and J48 is superior to others in the Breast Cancer dataset.

41

July 15, 2020(Wednsday) Room I 10:00-11:20

Learning Automata-based Fireworks Algorithm On AdaptiveAssigning Sparks

Zhang Junqi, Che Lei and Chen JianqingAbstract. Fireworks algorithm (FWA) is an emerging swarm intelligence inspired by the phenomenonof fireworks explosion. The numbers of sparks generated by fireworks have a great impact on thealgorithm performance. It is widely accepted that promising fireworks should generate more sparks.However, in many researches, the quality of a firework is judged only on its current fitness value. Thiswork proposes a Learning Automata-based Fireworks Algorithm (LA-FWA) introduced Learningautomata(LA) to assign sparks for a better algorithm performance. Sparks are assigned to fireworksaccording to a state probability vector, which is updated constantly based on feedbacks from anenvironment so that it accumulates historical information. The probability vector converges asthe search proceeds so that the local search ability of the LAFWA turns strong in the late searchstage. Experimental results performed on CEC2013 benchmark functions show that the LAFWAoutperforms several pioneering FWA variants.

A Novel Biogeography-based Optimization Algorithm withMomentum Migration and Taxonomic Mutation

Xinchao Zhao, Yisheng Ji and Junling HaoAbstract. Biogeography-based optimization (BBO) algorithm is not good at dealing with regionswhere function values change dramatically or barely. A novel biogeography-based optimizationalgorithm is proposed in this paper based on Momentum migration and taxonomic mutation. Themomentum item is added to the original migration operation of BBO. It makes the algorithmmore advantageous in dealing with regions where function values change dramatically or barely. Atthe same time, taxonomic mutation strategy divides the solutions into three categories: promisingclass, middle class and inferior class. Promising solutions do not take part in this mutationoperation. Solutions of middle class use balanced differential mutation, and inferior solutions adoptexploration-biased random mutation. This strategy further increases the diversity of population. Thesimulation experiments are carried out with different types of CEC2014 benchmark functions. Theproposed algorithm is compared with other algorithms and shows stronger global search ability, fasterconvergence speed and higher convergence accuracy.

Binary Pigeon-Inspired Optimization for Quadrotor SwarmFormation Control

Zhiqiang Zheng, Haibin Duan and Chen WeiAbstract. This paper proposes a binary pigeon-inspired optimization (BPIO) algorithm, for thequadrotor swarm formation control problem. The expected position is provided by the BPIO.Quadrotor moves to the position with control strategy, and the strategy is based on the proportionalintegral derivative (PID) control method. The BPIO algorithm which is based on pigeon-inspiredoptimization (PIO) algorithm can effectively solve the combination problem in the binary solutionspace. The BPIO keeps the fast convergence of the PIO, and can explore the space effectively at thesame time. The parameters to be optimized are encoded with binary bits. A special fitness functionis designed to avoid the happening of crash. The simulation experiment shows how the BPIO works.The results of simulation verify the feasibility and effectiveness of the BPIO to solve the swarmformation problem.

The Research of Flexible Scheduling of Workshop Based onArtificial Fish Swarm Algorithm and Knowledge Mining

Jieyang Peng, Jiahai Wang, Dongkun Wang, Andreas Kimmig and Jivka OvtcharovaAbstract. The Job Shop Scheduling problem is critical in the manufacturing industry. At present,the decision tree reasoning technique and data mining are often used in multi-objective optimizationresearch to solve flexible job shop scheduling issues. Unfortunately, when job shop scheduling

42


problems involve complex logic, it becomes difficult to implement data-driven automatic schedulingwithout human intervention. Based on the analysis of mass data and specialized knowledge in thescheduling domain, an ontology-based scheduling knowledge model and a method of knowledgerepresentation can be established. Considering the relationship between data mining and knowledge,this paper illustrates the acquisition process of scheduling rules. These scheduling rules were appliedto improve the initialization process of the artificial fish algorithm. Then, a scheduling experimentwas designed, the results of which show that the efficiency and accuracy of the algorithm has beenimproved. The desired uncertain information analysis, decision-making support for productionplanning and scheduling on the shop floor are provided and an adaptive scheduling algorithm forcomplex manufacturing systems is established by building a knowledge-based system.

July 15, 2020(Wednsday) Room II 10:00-11:20

A Modified Artificial Bee Colony Algorithm for SchedulingOptimization of Multi-Aisle AS/RS System

Xiaohui Yan, Felix T. S. Chan, Zhicong Zhang, Cixing Lv and Shuai LiAbstract. A modified artificial bee colony algorithm is proposed for solving the schedulingoptimization problem of multi-aisle automatic storage / retrieval system. The optimization modelof the problem is analyzed and founded, in which the sequence constraint of tasks and calculationof the number of aisles are more realistic. According to the features of the problem, the encodingand decoding strategies for solutions to MABC algorithm are redesigned. Probability selection-basedupdating method is also introduced to enhance the neighborhood search and preserve the goodfragments. The experimental results show that MABC can obtain better results than PSO and GAalgorithm, and is a competitive approach for AS/RS scheduling optimization.

Canine Algorithm for Node Disjoint PathsR Ananthalakshmi Ammal, P C Sajimon and Vinod Chandra S. S.

Abstract. Node Disjoint Paths (NDP) is one of the extensively studied Graph Theory problem. Inthis problem, the input is a directed n vertex graph and the set of source destination pair of vertices.The goal is to find the maximum number of paths connecting each such pair, so that such discoveredpaths are node-disjoint. In this paper, a novel Canine Inspired Algorithm is proposed which is abio-inspired one, based on the olfactory capabilities of canines in tracing and reaching a destination.Currently many of the existing algorithms try to identify disjoint paths in a linear manner, whereasthe Canine algorithm can be executed in a concurrent manner, depending on the number of caninesdeployed to find the disjoint paths. The time complexity of the algorithm is estimated to be .We hope that this algorithm finds many applications in problems related to various fields such ascommunication networks, scheduling and transportation and provides better results.

Research on PM2.5 Integrated Prediction Model Based onlasso-RF-GAM

Yan Peng, Tingxian Wu, Ziru Zhao and Haoxiang WeiAbstract. PM2.5 concentration is very difficult to predict, for it is the result of complex interactionsamong various factors. This paper combines the random forest-recursive feature elimination algorithmand lasso regression for joint feature selection, puts forward a PM2.5 concentration prediction modelbased on GAM. Firstly, the original data is standardized in the data input layer. Secondly, featureswere selected with RF-RFE and lasso regression algorithm in the feature selection layer. Meanwhile,weighted average method fused the two feature subsets to obtain the final subset, RF-lasso-T. Finally,the generalized additive models (GAM) is used to predict PM2.5 concentration on the RF-lasso-T.Simulated experiments show that feature selection allows GAM model to run more efficiently. Thedeviance explained by the model reaches 91.5%, which is higher than only using a subset of RF-RFE.This model also reveals the influence of various factors on PM2.5, which provides the decision-makingbasis for haze control.

43

An Evaluation Algorithm of the Importance of Network NodeBased on Community InfluenceGongzhen He, Junyong Luo and Meijuan Yin

Abstract. Identifying nodes in social networks that have great influence on information disseminationis of great significance for monitoring and guiding information dissemination. There are few methodsto study the influence of communities on social networks among the existing node importanceevaluation algorithms, and it is difficult to find nodes that promote information dissemination amongcommunities. In view of this reason, this paper proposes a node importance evaluation algorithmbased on community influence (abbreviated as IEBoCI algorithm), which evaluates the importanceof the nodes based on the influence degree of the nodes on the communities and the ability todisseminate information the communities to which the nodes are connected. This algorithm firstlycalculates the activation probability of nodes to other nodes, which is used to divide communitiesand evaluate influence. Secondly, the network is divided into communities based on LPA algorithm.Finally, the importance of the node is calculated by combining the influence of the community itselfand the influence of the node on the community. Experiments are carried out on real social networkdata and compared with other community-based methods to verify the effectiveness of the algorithm.

July 15, 2020(Wednsday) Room I 14:00-15:40

Target Tracking Algorithm based on Density ClusteringChen Jin Long, Zeng Qinghao and Qin Xingguo

Abstract. The traditional Siamese network-based target tracking algorithm needs to use theconvolution feature of the target to scan around the target location when predicting the location ofthe target in the next frame image, and perform similarity calculation to obtain the similarity scorematrix with the highest score. It is the next frame target position. The highest similarity score oftendoes not represent the precise target position of the target, which is often affected by the sliding stepsize during scanning. Aiming at this problem, this paper proposes a target tracking method based ondensity clustering. By combining the Siamese network to predict the next frame target position, andadding the target’s motion trajectory information, the direction of the target motion is given moreweight, the other directions are given a smaller weight, and finally the target position is predicted bythe density clustering method. The results show that the proposed algorithm effectively improves theaccuracy of the target location prediction of the Siamese network when tracking targets.

A Method for Localization and Classification of BreastUltrasound Tumors

Wanying Mo, Yuntao Zhu and Chaoyun WangAbstract. Ultrasound instruments are suitable for large-scale examination of breast tumors,especially for women from Asian whose glands are dense. However, ultrasound images have thelow contrast and resolution, blurred boundary and artifacts, which bring great difficulties to theinterpretation of the junior doctor. However, traditional methods of breast ultrasound tumorrecognition often use manually extracted features to gradually realize ROI region location and tumorclassification with low accuracy, poor robustness and weak universality. Deep learning is limited to thelocation of tumor ROI region or the classification of a given tumor ROI region. In this paper, YOLOV3algorithm is used for breast ultrasound tumor recognition, which could realize ROI localization andtumor classification at the same time. In addition, K-Means is optimized by K-Means++ andK-Mediods algorithm to generate anchor boxes of YOLOV3, and based on the Darknet-53 networkstructure of YOLOV3, ResNet and DenseNet are combined to design ResNet-DenseNet-Darknet-53.The proposed method is tested on the breast ultrasound tumor data set .Experiments show that theimproved YOLOV3 algorithm shows better detection results on multiple evaluation indicators.

44


Case Classification Processing and Analysis Method forRespiratory Belt DataJinlong Chen and Mengke Jiang

Abstract. Human respiratory signal is the important physiological indicator to reflect the physicalcondition. The respiratory belt, compared with the other human respiratory data measurementmethods, has the advantages of being portable, cheap, non-invasive, etc. However, it is unclear whichfeatures of the breathing data can effectively classify the normal/abnormal state of breathing state.To solve the problem, we proposed a novel approach based on long-short-term-memory (LSTM) andbreathing features of respiratory data. First, LSTM structure were used, then compared the resultwith the traditional method which extract the feature to experiment (in our paper which is RIE (ratioof inspiratory time to expiratory time)). In the end, a novel methodology proposed which combinedthe RIE feature with the LSTM structure. Experiment the three methods above using 342 normaland abnormal 24-hour breathing data, the results show that the third method has higher classificationaccuracy.

A Tool for Supporting the Evaluation of Active LearningActivities

Waraporn JirapanthongAbstract. Active learning becomes a strategical approach for an educational principle. The studentengagement become a wider concern. Many researches have been proposed to support the approach.However, one of issues is how to effectively evaluate the performance and progress of students’learning. Although, having student engagement in a classroom is vital, the evaluation of students’performance is more important. However, keeping up the details or records of students’ progressis a difficult task. We therefore propose a support for instructors to evaluate the performance oftheir students. In particular, a prototype tool is designed and developed in order to facilitate theevaluation of activities based on an active learning class. The tool also encompasses the web servicefor a function of face feature recognition. Two scenarios of active learning classrooms are createdin order to evaluate the prototype tool. We also plan to create a larger number of scenarios whichinvolve different class objectives. The results show that the tool can detect and determine studentswith high precision values. However, the prototype tool takes a long time to be processed dependingon the size and number of photos.

Inferring Candidate CircRNA-disease Associations by Bi-randomWalk Based on CircRNA Regulatory Similarity

Chunyan Fan, Xiujuan Lei and Ying TanAbstract. Identification of associations between circular RNAs (circRNA) and diseases has becomea hot topic, which is beneficial for researchers to understand the disease mechanism. However,traditional biological experiments are expensive and time-consuming. In this study, we proposeda novel method named BWHCDA, which applied bi-random walk algorithm on the heterogeneousnetwork for predicting circRNA-disease associations. First, circRNA regulatory similarity is measuredbased on circRNA-miRNA interactions, and circRNA similarity is calculated by the average ofcircRNA regulatory similarity and Gaussian interaction profiles (GIP) kernel similarity for circRNAs.Similarly, disease similarity is the mean of disease semantic similarity and GIP kernel similarityfor diseases. Then, the heterogeneous network is constructed by integrating circRNA network,disease network via circRNA-disease associations. Subsequently, the bi-random walk algorithm isimplemented on the heterogeneous network to predict circRNA-disease associations. Finally, weutilize leave-one-out cross validation and 10-fold cross validation frameworks to evaluate the predictionperformance of BWHCDA method and obtain AUC of 0.9334 and 0.8764+/-0.0038, respectively.Moreover, the predicted hsa-circ-0000519-gastric cancer association is analyzed. Results show thatBWHCDA could be an effective resource for clinical experimental guidance.

July 15, 2020(Wednsday) Room II 14:00-15:40

45

O-flocking: Optimized Flocking Model on AutonomousNavigation for Robotic Swarm

Li Ma, Weidong Bao, Xiaomin Zhu, Meng Wu, Yuan Wang, Yunxiang Ling and Wen ZhouAbstract. Flocking model has been widely used in robotic swarm control. However, the traditionalmodel still has some problems such as manually adjusted parameters, poor stability and lowadaptability when dealing with autonomous navigation tasks in large-scale groups and complexenvironments. Therefore, it is an important and meaningful research problem to automaticallygenerate Optimized Flocking model (O-flocking) with better performance and portability. Tosolve this problem, we design Comprehensive Flocking (C-flocking) model which can meet therequirements of formation keeping, collision avoidance of convex and non-convex obstacles anddirectional movement. At the same time, Genetic Optimization Framework for Flocking Model (GF)is proposed. The important parameters of C-flocking model are extracted as seeds to initialize thepopulation, and the offspring are generated through operations such as crossover and mutation. Theoffspring model is input into the experimental scene of autonomous navigation for robotic swarms,and the comprehensive fitness function value is obtained. The model with smallest value is selectedas the new seed to continue evolution repeatedly, which finally generates the O-flocking model. Theextended simulation experiments are carried out in more complex scenes, and the O-flocking andC-flocking are compared. Simulation results show that the O-flocking model can be migrated andapplied to large-scale and complex scenes, and its performance is better than that of C-flocking modelin most aspects.

A Parallel Evolutionary Algorithm with Value Decomposition forMulti-Agent Problems

Gao Li, Qiqi Duan and Yuhui ShiAbstract. Many real-world problems involve cooperation and/or competition among multiple agents.These problems often can be formulated as multi-agent problems. Recently, Reinforcement Learning(RL) has made significant progress on single-agent problems. However, multi-agent problems stillcannot be easily solved by traditional RL algorithms. First, the multi-agent environment is consideredas a non-stationary system. Second, most multi-agent environments only provide a shared teamreward as feedback. As a result, agents may not be able to learn proper cooperative or competitivebehaviors by traditional RL. Our algorithm adopts Evolution Strategies (ES) for optimizing policywhich is used to control agents and a value decomposition method for estimating proper fitnessfor each policy. Evolutionary Algorithm is considered as a promising alternative for signal-agentproblems. Owing to its simplicity, scalability, and efficiency on zeroth-order optimization, EAscan even outperform RLs on some tasks. In order to solve multi-agent problems by EA, a valuedecomposition method is used to decompose the team reward. Our method is parallel on multiplecores, which can speed up our algorithm significantly. We test our algorithm on two benchmarkingenvironments, and the experiment results show that our algorithm is better than traditional RL andother representative gradient-free methods.

Research on Sliding Mode Control of UnderwaterVehicle-manipulator System Based on an Exponential Approach

LawQirong Tang, Yang Hong, Zhenqiang Deng, Daopeng Jin and Yinghao Li

Abstract. To improve the performance of underwater vehicle-manipulator system (UVMS), which issubject to system uncertainties and time-varying external disturbances in trajectory tracking control,a sliding mode controller is proposed in this paper. Firstly, in order to reduce a influence of systemuncertainties and external disturbances, a sliding mode controller is designed based on an exponentialapproach law. Then the error asymptotic convergence of the trajectory tracking control is proven bythe Lyapunov-like function. Finally, the effectiveness of the sliding mode controller is verified by richsimulation. Results show that the designed controller can not only realize the coordination control ofUVMS accurately, but also can eliminate the chattering of control signal.

Multi-objective Combinatorial Generative AdversarialOptimization and Its Application in Crowdsensing

Yi-nan Guo, Jianjiao Ji, Ying Tan and Shi Cheng

46


Abstract. With the increasing of the decision variables in multi-objective combinatorial optimizationproblems, the traditional evolutionary algorithms perform worse due to the low efficiency forgenerating the offspring by a stochastic mechanism. To address the issue, a multi-objectivecombinatorial generative adversarial optimization method is proposed to make the algorithm capableof learning the implicit information embodied in the evolution process. After classifying the optimalnon-dominated solutions in the current generation as real data, the generative adversarial network(GAN) is trained by them, with the purpose of learning their distribution information. The Adamalgorithm that employs the adaptively learning rate for each parameter is introduced to update themain parameters of GAN. Following that, an offspring reproduction strategy is designed to form anew feasible solution from the decimal output of the generator. To further verify the rationality ofthe proposed method, it is applied to solve the participant selection problem of the crowdsensing andthe detailed offspring reproduction strategy is given. The experimental results for the crowdsensingsystems with various tasks and participants show that the proposed algorithm outperforms the othersin both convergence and distribution.

Multi-Objective Dynamic Scheduling Model of Flexible Job ShopBased on NSGAII Algorithm and Scroll Window Technology

Yingli Li and Jiahai WangAbstract. The production process is often accompanied by a lot of disturbances, which make itdifficult for flexible job shop to execute production according to the original job plan. It is necessaryto dynamically adjust the production plan according to real-time conditions. To this end, this paperproposes a multi-objective dynamic scheduling model. In this model, scroll window technology andNSGAII algorithm is adopted to adapt the dynamic production evironment. A specific chromosomeretention strategy and a variable objective selection mechanism are designed to ensure that theproposed model can select different objectives according to different disturbance events to solve theoptimal solution. Finally, a case test is used to verify the feasibility and effectiveness of the model.

47

Index (a=abstract c=chair cc=cochair)

Akhmedova, Shakhnaz, 21, 22, 38a, 41a

Akimenko, Tatyana, 21, 40a

Alberto, Pineres-Melo Marlon, 16, 26a

angulo, wilson rios, 16, 27a

Antonov, Maxim, 21, 40a

Araujo, Luiz Jonata Pires De, 21, 39a

Arseniev, Dmitriy, 21, 39a

Bacanin, Nebojsa, 21, 38a

Bao, Weidong, 24, 46a

Bezdan, Timea, 21, 38a

bonerge, omar, 18, 20, 31, 32a, 37a

Broneske, David, 21, 39a

Bu, Guan-Nan, 20, 35a

Bu, Hui, 19, 33a

Cao, Weifu, 17, 28a

Cesar, Morales-Ortega Roberto, 16, 26a

Chan, Felix T. S., 23, 43a

Chen, Jinlong, 24, 45a

Chen, Junfeng, 19, 20, 34, 35a

Cheng, Shi, 18, 19, 24, 30a, 33a, 46a

Chiba, Kazuhisa, 17, 28a

Clementina, Ospino-Mendoza Elisa, 16, 26a

Colpas, Paola Patricia Ariza, 16, 26a

Comas, Andres Gabriel Sanchez, 16, 26a

Contreras, Rodrigo, 16, 26a

Dai, Cai, 19, 34a

Darrab, Sadeq, 21, 22, 39a, 41a

Deng, Zhenqiang, 24, 46a

Duan, Haibin, 23, 42a

Duan, Qiqi, 18, 24, 30a, 46a

Ebrahimpour-Komleh, Hossein, 22, 40a

Enrique, Mendoza-Palechor Fabio, 16, 26a

Eugenia, Arrieta-Rodriguez, 16, 26a

Ezzeddine, Anna Bou, 21, 39a

Fan, Chunyan, 24, 45a

Fang, Hui, 22, 40a

Fu, Anbing, 20, 37a

Fu, Jia, 19, 35a

gaitan, mercedes, 17, 27a

Gaitan-Angulo, Mercedes, 16, 27a

Gan, Xiaobing, 17, 29a

Gao, Aiqing, 17, 29a

German, Lozano-Bernal, 16, 26a

Gimaeva, Susanna, 21, 39a

Grichshenko, Alexandr, 21, 39a

Grmanova, Gabriela, 21, 39a

Gulan, Maja, 19, 34a

Guo, Wenjing, 20, 37a

Guo, Yi-nan, 24, 46a

Gutierrez, Jenny Paola Lis, 17, 27a

gutierrez, melissa lis, 17, 27a

Hao, Junling, 23, 42a

Hatta, Taiki, 17, 28a

He, Gongzhen, 23, 44a

Henao, Carolina, 17, 27a

Hernandez, Jairo Coronado, 16, 27a

Holewik, Jakub, 21, 38a

Hong, Yang, 24, 46a

Hu, Ren-Yuan, 20, 35a

Hu, Ting, 20, 36a

Hu, Yang, 20, 35a

Huang, Miaojia, 17, 28a

Huang, Xin, 17, 29a

Iakovidis, Georgia Sovatzidi and Dimitris, 22,

41a

Ivutin, Alexey, 21, 38a

Ji, Jianjiao, 24, 46a

Ji, Yisheng, 23, 42a

Jiang, Jingzhou, 17, 29a

Jiang, Junjun, 19, 34a

Jiang, Lei, 20, 35a

Jiang, Mengke, 24, 45a

Jianqing, Chen, 23, 42a

Jin, Daopeng, 24, 46a

Jin, Jin, 20, 37a

48


Jirapanthong, Waraporn, 24, 45a

Jose, Caicedo-Ortiz, 16, 26a

Junior, Orides Morandin, 16, 26a

Junqi, Zhang, 23, 42a

Kanazaki, Masahiro, 17, 28a

Kimmig, Andreas, 23, 42a

Korovin, Iakov, 21, 22, 38a, 40a

Kostenko, Dmitri, 21, 39a

Larkin, Eugene, 21, 40a

Lei, Che, 23, 42a

Lei, Xiujuan, 24, 45a

Li, Gao, 24, 46a

Li, Hua, 19, 34a

Li, Jianjun, 19, 35a

Li, Shuai, 23, 43a

Li, Yinghao, 24, 46a

Li, Yingli, 25, 47a

Ling, Yunxiang, 24, 46a

Lis-Gutierrez, Jenny Paola, 16, 17, 27a

Lis-Gutierrez, Melissa, 16, 27a

Liu, Jian-Hua, 20, 35a

Liu, Ke, 20, 35a

Liu, Mingde, 18, 30a

Liu, Xiyao, 22, 40a

Loganathan, K, 18, 20, 31, 32a, 37a

Long, Chen Jin, 24, 44a

Lu, Mingli, 19, 33a

Luo, Junyong, 23, 44a

Luo, Yi-Xuan, 20, 35a

Luo, Yuxi, 17, 28a

Lv, Cixing, 23, 43a

Ma, Li, 24, 46a

Ma, Tao, 20, 36a

Margarita, Roca-Vides, 16, 26a

Mei, Yuan, 20, 36a

Melo, Marlon Alberto Pineres, 16, 26a

Mo, Wanying, 24, 44a

Mohammed, Siham A., 22, 41a

mojica, leonor, 17, 27a

Mousavirad, Seyed Jalaleddin, 22, 40a

Niu, Ben, 19, 34a

nino, diana urrego, 16, 27a

Noaman, Salah A., 22, 41a

Onufriev, Vadim, 21, 39a

ospino, luis ortiz, 18, 31a

Ou, Yikun, 17, 29a

Ovtcharova, Jivka, 23, 42a

Pan, Hongyue, 20, 35a

Patricia, Ariza-Colpas Paola, 16, 26a

Peng, Jieyang, 23, 42a

Peng, Wei, 20, 35a

Peng, Yan, 23, 43a

Phoa, Frederick Kin Hing, 19, 34a

Pinheiro, Rodrigo Lankaites, 21, 39a

Prasad, Sreedevi, 18, 32a

Privalov, Aleksandr, 21, 40a

Qinghao, Zeng, 24, 44a

Qiu, Zishan, 17, 28a

Qu, Liang, 18, 19, 30a, 33a

romero, alfonso, 16, 27a

romero, manuel, 17, 27a

Rong, Xin, 19, 35a

Rong, Yuxi, 20, 35a

Royert, Judith Martinez, 16, 26a

Rozinajova, Viera, 21, 39a

S, Dhanya, 18, 32a

S, Saju Sankar, 18, 30, 31a

S, Vinod Chandra S, 18, 30, 31a

S., Anand Hareendran S., Vinodchandra S.S. and

Saju Sankar, 18, 19, 32a

S., R Ananthalakshmi Ammal, P C Sajimon and Vinod

Chandra S., 23, 43a

S.S., Anand Hareendran S, Vinodchandra, 18,

32a

Saake, Gunter, 21, 22, 39a, 41a

Sagayaraj, A. Charles, 18, 31a

Saraiva, Rommel D., 21, 39a

Savenkov, Pavel, 21, 38a

Schaefer, Gerald, 21, 22, 38a, 40a

segovia, camilo, 16, 27a

Semenkin, Eugene, 22, 41a

49

Shen, Xiaolong, 19, 33a

Shen, Yang, 18, 19, 30a, 33a

Shi, Yuhui, 18, 19, 24, 30a, 33a, 46a

Shkodyrev, Vyacheslav, 21, 39a

Stanovov, Vladimir, 21, 22, 38a, 41a

Strumberger, Ivana, 21, 38a

Sumega, Miroslav, 21, 39a

Sun, Zhen, 19, 33a

Tamilvanan, k, 18, 20, 32a, 37a

Tan, Ying, 20, 24, 36a, 45, 46a

Tan, Yingsi, 17, 28a

Tang, Qirong, 24, 46a

Tapias, Belina Annery Herrera, 16, 26a

torres, adriana patricia gallego, 17, 27a

Tsai, Tzu-Chieh, 19, 34a

Tuba, Eva, 21, 38a

Tuba, Milan, 21, 38a

Tungom, Chia Emmanuel, 19, 34a

varela, noel, 18, 20, 31, 32a, 37a

Vargas-Garcıa, Cesar, 16, 27a

Viana, Monique, 16, 26a

viloria, amelec, 18, 20, 31, 32a, 37a

Wan, Liangpeng, 19, 34a

Wang, Chaoyun, 24, 44a

Wang, Dongkun, 23, 42a

Wang, Hong, 17, 29a

Wang, Jiahai, 23, 25, 42a, 47a

Wang, Xiaoling, 19, 35a

Wang, Yifeng, 19, 20, 34, 35a

Wang, Yuan, 24, 46a

Wei, Chen, 23, 42a

Wei, Haoxiang, 23, 43a

Wei, Wenhong, 20, 37a

Wu, Di, 19, 33a

Wu, Fangmin, 19, 20, 34, 35a

Wu, Meng, 24, 46a

Wu, Tingxian, 23, 43a

Wu, Xinzheng, 17, 28, 29a

Xiao, Baoyu, 17, 29a

Xingguo, Qin, 24, 44a

Xiong, Xiaojun, 17, 29a

Xu, Benlian, 19, 33a

Xue, Xingsi, 19, 20, 34, 35a

Yan, Xiaohui, 23, 43a

Yang, Chuan, 20, 37a

Yang, Gang, 20, 36a

Yang, Jian, 18, 19, 30a, 33a

Yang, Lichun, 20, 36a

Yang, Mingli Shi, Lianbo Ma and Guangming,

20, 37a

Yang, Yu, 19, 35a

Yao, Hanguang, 19, 34a

Yin, Meijuan, 23, 44a

Zhang, Dong-Yang, 20, 35a

Zhang, Yi, 20, 36, 37a

Zhang, Yuhui, 20, 37a

Zhang, Yun, 19, 34a

Zhang, Zhicong, 23, 43a

Zhao, Junfeng, 19, 33a

Zhao, Xinchao, 23, 42a

Zhao, Ziru, 23, 43a

Zheng, Ruiqi, 18, 30a

Zheng, Zhiqiang, 23, 42a

Zhou, Tianwei, 17, 28a

Zhou, Wen, 24, 46a

Zhu, Xiaomin, 24, 46a

Zhu, Yuntao, 24, 44a

50

Documents

Welcome Message from General Chairsdmbd2020.ic-si.org/program/program-icsi-dmbd2020.pdfthe ICSI-DMBD’2020 will provide you with such excellent program and schedule. The technical