[AISWorld] Newly published papers of JCSE (Mar. 2022)
JCSE office
office at kiise.org
Wed Mar 30 06:41:55 EDT 2022
Dear Colleague:
We are pleased to announce the release of a new issue of Journal of
Computing Science and Engineering (JCSE), published by the Korean Institute
of Information Scientists and Engineers (KIISE). KIISE is the largest
organization for computer scientists in Korea with over 4,000 active
members.
Journal of Computing Science and Engineering (JCSE) is a peer-reviewed
quarterly journal that publishes high-quality papers on all aspects of
computing science and engineering. JCSE aims to foster communication between
academia and industry within the rapidly evolving field of Computing Science
and Engineering. The journal is intended to promote problem-oriented
research that fuses academic and industrial expertise. The journal focuses
on emerging computer and information technologies including, but not limited
to, embedded computing, ubiquitous computing, convergence computing, green
computing, smart and intelligent computing, and human computing. JCSE
publishes original research contributions, surveys, and experimental studies
with scientific advances.
Please take a look at our new issue posted at http://jcse.kiise.org
<http://jcse.kiise.org/> . All the papers can be downloaded from the Web
page.
The contents of the latest issue of Journal of Computing Science and
Engineering (JCSE)
Official Publication of the Korean Institute of Information Scientists and
Engineers
Volume 16, Number 1, March 2022
pISSN: 1976-4677
eISSN: 2093-8020
* JCSE web page: http://jcse.kiise.org
* e-submission: http://mc.manuscriptcentral.com/jcse
Editor in Chief: Insup Lee (University of Pennsylvania)
Il-Yeol Song (Drexel University)
Jong C. Park (KAIST)
Taewhan Kim (Seoul National University)
JCSE, vol. 16, no. 1, March 2022
[Paper One]
- Title: Deep and Statistical-Based Methods for Alzheimer's Disease
Detection: A Survey
- Authors: Marwa Zaabi, Nadia Smaoui, Walid Hariri, and Houda Derbel
- Keyword: Alzheimer's disease; Statistical methods; Deep learning methods;
Segmentation methods
- Abstract
Detection of Alzheimer's disease (AD) is one of the most potent and daunting
activities in the processing of medical imagery. The survey of recent AD
detection techniques in the last 10 years is described in this paper. The AD
detection process involves various steps, namely preprocessing, feature
extraction, feature selection, dimensionality reduction, segmentation and
classification. In this study, we reviewed the latest findings and possible
patterns as well as their main contributions. Different types of AD
detection techniques are also discussed. Based on the applied algorithms and
methods, and the evaluated databases (e.g., ADNI and OASIS), the
performances of the most relevant AD detection techniques are compared and
discussed.
To obtain a copy of the entire article, click on the link below.
JCSE, vol. 16, no. 1, pp.1-13
<http://jcse.kiise.org/PublishedPaper/year_abstract.asp?idx=396&page_url=Cur
rent_Issues>
[Paper Two]
- Title: Improvement in Object Detection Using Multi-Scale RoI Pooling and
Feature Pyramid Network
- Authors: Seungtae Nam and Daeho Lee
- Keyword: Region of interest pooling; Feature pyramid network; Object
detection; Deep learning
- Abstract
The feature pyramid network (FPN) enhances the localization accuracy and
detection performance of small objects using multiple scales of the
features. FPN adopts lateral connections and a top-down pathway to make
low-level features semantically more meaningful. However, it uses only
single-scale features to pool regions of interest (RoIs) when detecting
objects. In this study, we showed that single-scale RoI pooling may not be
the best solution for accurate localization and proposed multi-scale RoI
pooling to improve the minor drawbacks of the FPN. The proposed method pools
RoIs from three feature levels and concatenates the pooled features to
detect objects. Thus, the FPN with multi-scale RoI pooling, called FPN+,
detects objects by taking into account all information scattered across
three feature levels. FPN+ improved the FPN by 2.81 and 1.1 points in
COCO-style average precision (AP) when tested on PASCAL VOC 2007 test and
COCO 2017 validation datasets, respectively.
To obtain a copy of the entire article, click on the link below.
JCSE, vol. 16, no. 1, pp.14-24
<http://jcse.kiise.org/PublishedPaper/year_abstract.asp?idx=397&page_url=Cur
rent_Issues>
[Paper Three]
- Title: Automatic Modulation Recognition Using Minimum-Phase Reconstruction
Coefficients and Feed-Forward Neural Network
- Authors: Sunday Ajala, Emmanuel Adetiba, Oluwaseun T. Ajayi, Abdultaofeek
Abayomi, Anabi Hilary Kelechi, Joke A. Badejo, Sibusiso Moyo, and Murimo
Bethel Mutanga
- Keyword: Cognitive radio; Cepstrum analysis; GNU Radio; Modulation
schemes; MPRC; RCC
- Abstract
Identification of signal waveforms is highly critical in 5G communications
and other state-of-the-art radio technologies such as cognitive radios. For
instance, to achieve efficient demodulation and spectrum sensing, cognitive
radios need to implement automatic modulation recognition (AMR) of detected
signals. Although many works have been reported in the literature on the
subject, most of them have mainly focused on the additive white Gaussian
noise (AWGN) channel. However, addressing the AWGN channel, only, does not
sufficiently emulate real-time wireless communications. In this paper, we
created datasets of six modulation schemes in GNU Radio. Wireless signal
impairment issues such as center frequency offset, sample rate offset, AWGN,
and multipath fading effects were applied for the dataset creation.
Afterward, we developed AMR models by training different artificial neural
network (ANN) architectures using real cepstrum coefficients (RCC), and
minimum-phase reconstruction coefficients (MPRC) extracted from the created
signals. Between these two features, MPRC features have the best
performance, and the ANN architecture with Levenberg-Marquardt learning
algorithm, as well as logsig and purelin activation functions in the hidden
and output layers, respectively, gave the best performance of 98.7%
accuracy, 100% sensitivity, and 99.33% specificity when compared with other
algorithms. This model can be leveraged in cognitive radio for spectrum
sensing and automatic selection of signal demodulators.
To obtain a copy of the entire article, click on the link below.
JCSE, vol. 16, no. 1, pp.25-42
<http://jcse.kiise.org/PublishedPaper/year_abstract.asp?idx=398&page_url=Cur
rent_Issues>
[Paper Four]
- Title: Domain Transformation for PMMA-Inserted Vertebral Body Segmentation
- Authors: Minyoung Park and Jinah Park
- Keyword: Computer vision; Medical image segmentation; Domain
transformation
- Abstract
Learning-based medical image segmentation has been advanced with the
collection of datasets and various training methodologies. In this work, we
target bone cement (polymethylmethacrylate [PMMA]) inserted vertebral body
segmentation, where the target dataset was relatively scarce, compared to a
large-scale dataset for the regular vertebra segmentation task. We presented
a novel domain transformation framework, where a large-scale training set
for our target task was generated from the existing dataset of a different
domain. We proposed two main components: label translation and image
translation. Label translation was proposed to filter out unnecessary
regions in a segmentation map for our target task. In addition to label
translation, image translation was proposed to virtually generate
PMMA-inserted images from the original data. The synthesized training set by
our method successfully simulated the data distribution of the target task;
therefore a clear performance improvement was observed by the dataset. By
extensive experiments, we showed that our method outperformed baseline
methods in terms of segmentation performance. In addition, a more accurate
shape and volume of a bone were measured by our method, which satisfied the
medical purpose of segmentation.
To obtain a copy of the entire article, click on the link below.
JCSE, vol. 16, no. 1, pp.43-51
<http://jcse.kiise.org/PublishedPaper/year_abstract.asp?idx=399&page_url=Cur
rent_Issues>
[Paper Five]
- Title: Stable Federated Learning with Dataset CondensationForward Neural
Network
- Authors: Seong-Woong Kim and Dong-Wan Choi
- Keyword: Deep learning; Federated learning; Dataset compression; Class
imbalance
- Abstract
Federated learning (FL) is a new machine learning paradigm, where multiple
clients learn their local models to collaboratively integrate into a single
global model. Unlike centralized learning, the global model being integrated
cannot be tested in FL as the server does not collect any data samples,
further, the global model is often sent back and immediately applied to
clients even at the middle of training such as Gboard. Therefore, if the
performance of the global model is not stable, which is, unfortunately, the
case in many FL scenarios with non-IID data, clients can be provided with an
inaccurate model. This paper explores the main reason for this training
instability of FL, that is, what we call temporary imbalance that happens
across rounds, leading to loss of knowledge from previous rounds. To solve
this problem, we propose a dataset condensation method to summarize the
local data for each client without compromising on privacy. The condensed
data are transmitted to the server with the local model and utilized by the
server to ensure stable and consistent performance of the global model.
Experimental results show that the global model not only achieves training
stability but also exhibits a fast convergence speed.
To obtain a copy of the entire article, click on the link below.
JCSE, vol. 16, no. 1, pp.52-62
<http://jcse.kiise.org/PublishedPaper/year_abstract.asp?idx=400&page_url=Cur
rent_Issues>
[Call For Papers]
Journal of Computing Science and Engineering (JCSE), published by the Korean
Institute of Information Scientists and Engineers (KIISE) is devoted to the
timely dissemination of novel results and discussions on all aspects of
computing science and engineering, divided into Foundations, Software &
Applications, and Systems & Architecture. Papers are solicited in all areas
of computing science and engineering. See JCSE home page at
http://jcse.kiise.org <http://jcse.kiise.org/> for the subareas.
The journal publishes regularly submitted papers, invited papers, selected
best papers from reputable conferences and workshops, and thematic issues
that address hot research topics. Potential authors are invited to submit
their manuscripts electronically, prepared in PDF files, through
<http://mc.manuscriptcentral.com/jcse> http://mc.manuscriptcentral.com/jcse,
where ScholarOne is used for on-line submission and review. Authors are
especially encouraged to submit papers of around 10 but not more than 30
double-spaced pages in twelve point type. The corresponding author's full
postal and e-mail addresses, telephone and FAX numbers as well as current
affiliation information must be given on the manuscript. Further inquiries
are welcome at JCSE Editorial Office, <mailto:office at kiise.org>
office at kiise.org (phone: +82-2-588-9240; FAX: +82-2-521-1352).
More information about the AISWorld
mailing list