Professional Documents
Culture Documents
Neural Information Processing 27th International Conference ICONIP 2020 Bangkok Thailand November 23 27 2020 Proceedings Part III Haiqin Yang
Neural Information Processing 27th International Conference ICONIP 2020 Bangkok Thailand November 23 27 2020 Proceedings Part III Haiqin Yang
Neural Information Processing 27th International Conference ICONIP 2020 Bangkok Thailand November 23 27 2020 Proceedings Part III Haiqin Yang
https://textbookfull.com/product/neural-information-
processing-27th-international-conference-iconip-2020-bangkok-
thailand-november-23-27-2020-proceedings-part-i-haiqin-yang/
https://textbookfull.com/product/neural-information-
processing-27th-international-conference-iconip-2020-bangkok-
thailand-november-23-27-2020-proceedings-part-ii-haiqin-yang/
https://textbookfull.com/product/neural-information-
processing-27th-international-conference-iconip-2020-bangkok-
thailand-november-18-22-2020-proceedings-part-iv-haiqin-yang/
https://textbookfull.com/product/neural-information-
processing-27th-international-conference-iconip-2020-bangkok-
thailand-november-18-22-2020-proceedings-part-v-haiqin-yang/
Neural Information Processing: 25th International
Conference, ICONIP 2018, Siem Reap, Cambodia, December
13–16, 2018, Proceedings, Part III Long Cheng
https://textbookfull.com/product/neural-information-
processing-25th-international-conference-iconip-2018-siem-reap-
cambodia-december-13-16-2018-proceedings-part-iii-long-cheng/
https://textbookfull.com/product/computer-networks-27th-
international-conference-cn-2020-gdansk-poland-
june-23-24-2020-proceedings-piotr-gaj/
https://textbookfull.com/product/cooperative-design-
visualization-and-engineering-17th-international-conference-
cdve-2020-bangkok-thailand-october-25-28-2020-proceedings-yuhua-
luo/
Neural
Information Processing
27th International Conference, ICONIP 2020
Bangkok, Thailand, November 23–27, 2020
Proceedings, Part III
Lecture Notes in Computer Science 12534
Founding Editors
Gerhard Goos
Karlsruhe Institute of Technology, Karlsruhe, Germany
Juris Hartmanis
Cornell University, Ithaca, NY, USA
Neural
Information Processing
27th International Conference, ICONIP 2020
Bangkok, Thailand, November 23–27, 2020
Proceedings, Part III
123
Editors
Haiqin Yang Kitsuchart Pasupa
Department of AI Faculty of Information Technology
Ping An Life King Mongkut's Institute
Shenzhen, China of Technology Ladkrabang
Bangkok, Thailand
Andrew Chi-Sing Leung
City University of Hong Kong James T. Kwok
Kowloon, Hong Kong Department of Computer Science
and Engineering
Jonathan H. Chan Hong Kong University of Science
School of Information Technology and Technology
King Mongkut’s University Hong Kong, Hong Kong
of Technology Thonburi
Bangkok, Thailand Irwin King
The Chinese University of Hong Kong
New Territories, Hong Kong
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
This book is a part of the five-volume proceedings of the 27th International Conference
on Neural Information Processing (ICONIP 2020), held during November 18–22,
2020. The conference aims to provide a leading international forum for researchers,
scientists, and industry professionals who are working in neuroscience, neural net-
works, deep learning, and related fields to share their new ideas, progresses, and
achievements. Due to the outbreak of COVID-19, this year’s conference, which was
supposed to be held in Bangkok, Thailand, was organized as fully virtual conference.
The research program of this year’s edition consists of four main categories, Theory
and Algorithms, Computational and Cognitive Neurosciences, Human-Centered
Computing, and Applications, for refereed research papers with nine special sessions
and one workshop. The research tracks attracted submissions from 1,083 distinct
authors from 44 countries. All the submissions were rigorously reviewed by the con-
ference Program Committee (PC) comprising 84 senior PC members and 367 PC
members. A total of 1,351 reviews were provided, with each submission receiving at
least 2 reviews, and some papers receiving 3 or more reviews. This year, we also
provided rebuttals for authors to address the errors that exist in the review comments.
Meta-reviews were provided with consideration of both authors’ rebuttal and review-
ers’ comments. Finally, we accepted 187 (30.25%) of the 618 full papers that were sent
out for review in three volumes of Springer’s series of Lecture Notes in Computer
Science (LNCS) and 189 (30.58%) of the 618 in two volumes of Springer’s series of
Communications in Computer and Information Science (CCIS).
We would like to take this opportunity to thank all the authors for submitting their
papers to our conference, and the senior PC members, PC members, as well as all the
Organizing Committee members for their hard work. We hope you enjoyed the
research program at the conference.
Honorary Chairs
Jonathan Chan King Mongkut’s University of Technology Thonburi,
Thailand
Irwin King Chinese University of Hong Kong, Hong Kong
General Chairs
Andrew Chi-Sing Leung City University of Hong Kong, Hong Kong
James T. Kwok Hong Kong University of Science and Technology,
Hong Kong
Program Chairs
Haiqin Yang Ping An Life, China
Kitsuchart Pasupa King Mongkut’s Institute of Technology Ladkrabang,
Thailand
Finance Chairs
Vajirasak Vanijja King Mongkut’s University of Technology Thonburi,
Thailand
Seiichi Ozawa Kobe University, Japan
Tutorial Chairs
Zenglin Xu Harbin Institute of Technology, China
Jing Li Hong Kong Polytechnic University, Hong Kong
viii Organization
Proceedings Chairs
Xinyi Le Shanghai Jiao Tong University, China
Jinchang Ren University of Strathclyde, UK
Publicity Chairs
Zeng-Guang Hou Chinese Academy of Sciences, China
Ricky Ka-Chun Wong City University of Hong Kong, Hong Kong
Program Committee
Muideen Adegoke City University of Hong Kong, Hong Kong
Sheraz Ahmed German Research Center for Artificial Intelligence,
Germany
Shotaro Akaho National Institute of Advanced Industrial Science
and Technology, Japan
Sheeraz Akram University of Pittsburgh, USA
Abdulrazak Alhababi Universiti Malaysia Sarawak, Malaysia
Muhamad Erza Aminanto University of Indonesia, Indonesia
Marco Anisetti University of Milan, Italy
Sajid Anwar Institute of Management Sciences, Pakistan
Muhammad Awais COMSATS University Islamabad, Pakistan
Affan Baba University of Technology Sydney, Australia
Boris Bacic Auckland University of Technology, New Zealand
Mubasher Baig National University of Computer and Emerging
Sciences, Pakistan
Tao Ban National Information Security Research Center, Japan
Sang Woo Ban Dongguk University, South Korea
Kasun Bandara Monash University, Australia
David Bong Universiti Malaysia Sarawak, Malaysia
George Cabral Rural Federal University of Pernambuco, Brazil
Anne Canuto Federal University of Rio Grande do Norte, Brazil
Zehong Cao University of Tasmania, Australia
Jonathan Chan King Mongkut’s University of Technology Thonburi,
Thailand
Guoqing Chao Singapore Management University, Singapore
Hongxu Chen University of Technology Sydney, Australia
Ziran Chen Bohai University, China
Xiaofeng Chen Chongqing Jiaotong University, China
Xu Chen Shanghai Jiao Tong University, China
He Chen Hebei University of Technology, China
Junjie Chen Inner Mongolia University, China
Mulin Chen Northwestern Polytechnical University, China
Junying Chen South China University of Technology, China
Organization xi
Biomedical Information
Recommender Systems
GFEN: Graph Feature Extract Network for Click-Through Rate Prediction . . . 444
Mei Yu, Chengchang Zhen, Ruiguo Yu, Xuewei Li, Tianyi Xu,
Mankun Zhao, Hongwei Liu, Jian Yu, and Xuyuan Dong
LHRM: A LBS Based Heterogeneous Relations Model for User Cold Start
Recommendation in Online Travel Platform . . . . . . . . . . . . . . . . . . . . . . . . 479
Ziyi Wang, Wendong Xiao, Yu Li, Zulong Chen, and Zhi Jiang
SpringNet: Transformer and Spring DTW for Time Series Forecasting. . . . . . 616
Yang Lin, Irena Koprinska, and Mashud Rana
1 Introduction
Neuroblastoma is the most common cancer diagnosed in children in the first year
of life and accounts for nearly 15% of deaths in children due to cancer [1,2]. Neu-
roblast tumours evolve from immature neuroblasts in the sympathetic nervous
system during the embryonic, fetal or postnatal stage in children. The disease
c Springer Nature Switzerland AG 2020
H. Yang et al. (Eds.): ICONIP 2020, LNCS 12534, pp. 3–14, 2020.
https://doi.org/10.1007/978-3-030-63836-8_1
4 A. Panta et al.
spreads typically through bone, bone marrow and the liver, and the tumour
can be spotted as mass lesions in areas like neck, chest, abdomen and pelvis
[1,2]. Cellular heterogeneity is one of the distinctive features of neuroblastoma
[3]. As a result of this feature, neuroblastic tumours show unexpected clinical
behaviour, which includes spontaneous regression and aggressive progression. As
such, it is common to see genetic materials achieve gains and loss rapidly when
neuroblastoma is evolving.
Pathologists use the Shimada system to classify whether a tumour is
favourable or unfavourable, which is considered the gold standard in neuroblas-
toma classification [3]. The Shimada system considers three key factors, which
are: age of the patient, the category of the Neuroblast tumour and the Mitosis-
Karyorrhexis index (MKI) [3]. To categorise neuroblast tumours, pathologists
examine thin tissues using optical microscopes in different magnifications. While
identifying the category of tumours, pathologists use several morphological fea-
tures such as the presence of neuropil, cellularity, nuclear size and shape [3]. How-
ever, due to the complex and heterogeneous nature of neuroblastoma, pathol-
ogists can get misleading results. The use of machine learning techniques for
feature extraction can reveal information and relationships not visible to the
human eye. Moreover, the use of Computer-Aided Design (CAD) systems in the
health sector offer benefits such as improvement of the overall speed and qual-
ity of the diagnosis process by eliminating human fatigue, acting as a tool for
the second opinion and assisting with the shortage of medical experts [3]. These
benefits are especially significant for neuroblastoma as the age of a patient at
the time of diagnosis is vital for the prognosis outcome.
There is a lack of significant exploration of the use of machine learning tech-
niques in for classification of Nuroblastoma types. One of the recent work was
done by S. Gheisari et al. [5–7] who classified neuroblastoma histopathological
images into five categories which were: undifferentiated neuroblastoma, gan-
glioneuroblastoma, ganglioneuroma, poorly-differentiated neuroblastoma, and
differentiating neuroblastoma. A range of low level and high-level feature extrac-
tion techniques were used in previous research to achieve accuracy of around 84%.
From a medical perspective, more accurate results would be desirable to increase
confidence and improve the chances of computer-based systems being used to
assist experts. Furthermore, [5–7] also identified that there was a high degree of
misclassification between poorly-differentiated and differentiating neuroblastoma
classes. From a biological perspective, these misclassifications are significant as
they can result in patients being overtreated or undertreated. Hence there is
keen interest in improving existing methods.
This paper aims to improve the previous work performed by S. Gheisari
et al. [5–7], where neuroblastoma images were successfully classified into five
categories. The dataset used for this work is the same as used in [5–7] which was
gathered from The Tumour Bank at Kids Research at The Children’s Hospital
at Westmead. Through the exploration of previously used feature extraction
methods and existing data optimisation techniques, this study aims to improve
the overall accuracy metrics achieved in previous neuroblastoma research. The
contributions made by this study is summarised below:
Classifying Neuroblastoma Images 5
1. Explored the impact of different feature extraction and machine learning tech-
niques on the performance for classification of neuroblastoma histopatholog-
ical images.
2. Evaluated the effects of combining outcomes from different feature extraction
techniques on performance metrics.
3. Explored the effects of data engineering techniques such as resampling the
given dataset on the overall performance.
4. Improved overall accuracy metrics through the use of these techniques
The rest of the paper is as follows: Sect. 2 discusses the methodology used.
Section 3 shows the results. Section 4 presents the discussion and Sect. 5 provides
a conclusion.
2 Methodology
To effectively conduct the intended research, this work follows a structure com-
monly used for the application of machine learning in the medical informatics
domain. The first step involves relevant data needs to be collected, or exist-
ing dataset needs to be selected to conduct the experiments. Collecting medical
image is a complex process which involves taking tissue samples from high resolu-
tion microscopes. It would also require expert medical professionals to categorize
the data so that models can be trained. So, the collection of raw data was con-
sidered out of scope, and an existing neuroblastoma dataset was used to conduct
the experiments. The next step involves feature engineering, where the aim is to
extract relevant features from available image data. Multiple feature and data
engineering techniques are explored in work conducted. Then machine learning
classification algorithms are implemented for classification. In this study, SVM
classification is used for uniform comparison as a greater focus on the feature
extraction and data engineering methods. Finally, the designed models are tested
and evaluated using 5-fold cross-validation, accuracy, precision, recall and F-1
score.
2.1 Dataset
The dataset used for this study is the same as used by S. Gheisari et al. in [5–7].
The dataset was gathered from The Tumour Bank at Kids Research at The Chil-
dren’s Hospital at Westmead and is the most comprehensive available dataset
for research in neuroblastoma. There is no publicly available dataset for neu-
roblastoma research, so the authors granted access to this dataset. The dataset
contains 1043 images gathered from stained tissue biopsy slides of 125 patients.
The tissue slides were scanned by using a software called Imagescope under 40x
magnification. Each image was cropped to include 300 by 300 pixels to provide
a balance between achieving a reasonable computational time and preserving
critical information in each image. It was also ensured that each cropped image
contained areas that best represent each category of neuroblastoma. This is a
6 A. Panta et al.
slight limitation of the dataset because it does not directly mimic a real-world
scenario where data can have noise points. Expert pathologists classified images
in the dataset according to the Shimada System where the categories included:
poorly differentiated neuroblastoma, differentiating neuroblastoma, undifferen-
tiating neuroblastoma, ganglioneuroma and ganglioneuroblastoma.
Pages.
Avertissement I
Journal d’Eugénie de Guérin 1
I (15 novembre 1834-13 avril 1835) 3
II (14 avril-5 décembre 1835) 59
III (mars-mai 1836) 105
IV (mai-juin 1837) 119
V (26 janvier-19 février 1838) 143
VI (19 février-3 mai 1838) 161
VII (3 mai-29 septembre 1838) 199
VIII (10 avril-25 mai 1839) 247
IX (21 juillet 1839-9 janvier 1840) 275
X (9 janvier-19 juillet 1840) 327
XI (26 juillet-29 août 1840) 391
XII (1er novembre-31 décembre 1840) 409
Fragments 419
MADAME SWETCHINE
JOURNAL DE SA CONVERSION, MÉDITATIONS ET PRIÈRES. 1
volume in-8o 7 fr. 50
Le même ouvrage, 1 vol. in-12 3 fr. 50
VIE ET PENSÉES DE Mme SWETCHINE, publiées par M. le Comte
de Falloux, de l’Académie française, 2 vol. in-8o 15 fr.
Le même ouvrage, 2 vol. in-12 7 fr.
LETTRES DE Mme SWETCHINE, publiées par M. le Comte de
Falloux. 2 vol. in-8o 15 fr.
Le même ouvrage, 2 vol. in-12 7 fr.
CORRESPONDANCE DU R. P. LACORDAIRE ET DE Mme
SWETCHINE, publiée par M. le Comte de Falloux. 1 fort vol. in-
8o 7 fr. 50
Le même ouvrage, 1 vol. in-12 4 fr.
MADAME LA PRINCESSE DE BROGLIE
LES VERTUS CHRÉTIENNES, expliquées par des récits tirés de la
Vie des Saints. 2 beaux vol. in-12 illustrés 6 fr.
MADAME LA MARQUISE DE SAINTE-AULAIRE
LA CHANSON D’ANTIOCHE, composée par Richard le Pèlerin et
Graindor, de Douai, et traduite. 1 vol. in-12 3 fr. 50
MADAME LA COMTESSE D’***, NÉE DE SÉGUR
LA REINE MARIE LECKZINSKA. Étude historique. 1 vol. in-12. 3 fr.
MADAME A. TASTU
POÉSIES COMPLÈTES, nouvelle et très-jolie édition illustrée de
vignettes de Johannot. 1 fort vol. in-12 3 fr. 50
L’ÉDUCATION MATERNELLE, simples leçons d’une mère à ses
enfants sur la lecture, l’écriture, la grammaire, etc. 1 vol. grand in-
8o illustré 15 fr.
MADAME GUILLON-VIARDOT
CINQ ANNÉES DE LA VIE DES JEUNES FILLES. (L’entrée dans le
monde.) 1 joli vol. in-12 3 fr. 50
MADAME PENQUER
LES CHANTS DU FOYER, Poésies, 3e édit. 1 vol. in-12 3 fr. 50
MADAME LA VICOMTESSE DE DAX
L’AMOUR ET LA FEMME. 1 vol. in-12 2 fr.
MADAME ***
LA VIE DE VILLAGE EN ANGLETERRE, ou Souvenirs d’un Exilé,
par l’auteur de l’Étude sur Channing. 2e édit. 1 vol. in-12. 3 fr. 50
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside
the United States, check the laws of your country in addition to
the terms of this agreement before downloading, copying,
displaying, performing, distributing or creating derivative works
based on this work or any other Project Gutenberg™ work. The
Foundation makes no representations concerning the copyright
status of any work in any country other than the United States.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if
you provide access to or distribute copies of a Project
Gutenberg™ work in a format other than “Plain Vanilla ASCII” or
other format used in the official version posted on the official
Project Gutenberg™ website (www.gutenberg.org), you must, at
no additional cost, fee or expense to the user, provide a copy, a
means of exporting a copy, or a means of obtaining a copy upon
request, of the work in its original “Plain Vanilla ASCII” or other
form. Any alternate format must include the full Project
Gutenberg™ License as specified in paragraph 1.E.1.
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information