Forthcoming Articles

These articles have been peer-reviewed and accepted for publication in JICT, but are pending final changes, are not yet published and may not appear here in their final order of publication until they are assigned to issues. Therefore, the content conforms to our standards but the presentation (e.g. typesetting and proof-reading) is not necessarily up to the JICT standard. Additionally, titles, authors, abstracts and keywords may change before publication.

  
ENHANCED ROBUST UNIVARIATE CLASSIFICATION METHODS FOR SOLVING OUTLIERS AND OVERFITTING PROBLEMS
1Friday Zinzendoff Okwonu, 2Nor Aishah Ahad, 2Hashibah Hamid, 3Nora Muda & 4Olimjon S. Sharipov
1Department of Mathematics, Faculty of Science, Delta State University, Abraka, Nigeria
2School of Quantitative Sciences, College of Arts and Sciences, Universiti Utara Malaysia, Sintok Kedah
3School of Mathematical Sciences, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, Bangi Selangor
4Department of Statistics, School of Mathematics, National University of Uzbekistan named after Mirzo Ulugbek ,Tashkent, Uzbekistan
okwonufz@delsu.edu.ng; aishah@uum.edu.my; hashibah@uum.edu.my; noramuda@ukm.edu.my; osharipov@mail.ru
 

ABSTRACT

The robustness of some classical univariate classifiers is hampered if the data are contaminated. Overfitting is another hiccup when the data sets are uncontaminated with a considerable sample size. The performance of the classification models can be easily biased by the outliers’ problems, of which the constructed model tends to be overfitted. Previous studies often used the Bayes Classifier (BC) and the Predictive Classifier (PC) to address two groups of univariate classification problems. Unfortunately for substantial large sample sizes and uncontaminated data, the BC method overfits when the Optimal Probability of Exact Classification (OPEC) is used as an evaluation benchmark. Meanwhile, for small sample sizes, the BC and PC methods are extremely susceptible to outliers. To overcome these two problems, we proposed two methods: the Smart Univariate Classifier (SUC) and the hybrid classifier. The latter is a combination of the SUC and the BC methods, known as the Smart Univariate Bayes Classifier (SUBC). The performance of the new classification methods was evaluated and compared with the conventional BC and PC methods using the OPEC as a benchmark value. To validate the performance of these classification methods, the Probability of Exact Classification (PEC) was compared with the OPEC value. The results showed that the proposed methods outperformed the conventional BC and PC methods based on the real data sets applied. Numerical results also revealed that the SUC method could solve the overfitting problem. The results further indicated that the two proposed methods were robust against outliers. Therefore, these new methods are helpful when practitioners are confronted with overfitting and data contamination problems.

Keywords: Bayes; Predictive; Outliers; Overfitting; Classification.
 

 
IT'S COOL TO BE HEALTHY! THE EFFECT OF PERCEIVED COOLNESS ON THE ADOPTION OF FITNESS BANDS AND HEALTH BEHAVIOUR
Nurdayana Mohamad Noor, Izzal Asnira Zolkepli, & Bahiyah Omar 
School of Communication, Universiti Sains Malaysia, Malaysia
dayananoor@student.usm.my; izzalasnira@usm.my* bahiyah@usm.my
*corresponding author
 
ABSTRACT

Contemporary technology success is frequently associated with the competitive advantage of being cool. A fitness band is one of the smart wearable devices promoting health behaviours, which is one of the cool lifestyle trends in modern societies. Although past research established the profound effects of coolness on user technology acceptance, the influencing role in fostering health behaviour remained obscure. To bridge the existing literature gap, the current study aims to examine the perception of coolness as a higher-order construct with multiple dimensions, namely originality, attractiveness, and sub-cultural appeals by investigating the direct effect on fitness band adoption and indirect influence on users’ health behaviour. An online survey was conducted on 280 fitness band users, and the data was subsequently analysed via the Partial Least Squares -Structural Equation Modeling (PLS-SEM). The study results demonstrated that the perceived coolness of fitness bands significantly affected users’ device adoption levels, which subsequently influenced personal health behaviour. This study thus contributed to health communication research by testing the coolness concept and developing the diffusion-innovation framework from current human-computer interaction literature. The findings would guide future developers of fitness bands to emphasise the coolness functions for higher degrees of adoption and positive impact on society.

Keywords: smart wearable device, ubiquitous computing, health behaviour, technology adoption, human-computer interaction, user experience
 

 
TIME DISTRIBUTED ATTENTION LAYERED CONVOLUTION NEURAL NETWORK WITH ENSEMBLE LEARNING USING RANDOM FOREST CLASSIFIER FOR SPEECH EMOTION RECOGNITION
*1Y.Bhanusree, 2 S.Srinivas Kumar, & 3A. Koteswara Rao
1Department of CSE, Jawaharlal Nehru Technological University Kakinada, India,
2Department of ECE, Jawaharlal Nehru Technological University Kakinada, India,
Department of CSE, Kalasalingam Academy of Research and Education Tamilnadu, India
 
 
ABSTRACT
 
Speech Emotion Detection (SER) is a field of identifying human emotions from human speech utterances. Human speech utterances are a combination of linguistic and nonlinguistic information. Non-linguistic SER provides a generalized solution in human–computer interaction applications as it overcomes the language barrier. Machine learning and deep learning techniques were previously proposed for classifying emotions using handpicked features. To achieve effective and generalized SER, feature extraction can be performed using deep neural networks and ensemble learning for classification. The proposed model employed a time distributed attention-layered convolution neural network (TDACNN) for extracting spatiotemporal features at the first stage and a random forest (RF) classifier, which is an ensemble classifier for efficient and generalized classification of emotions, at the second stage. The proposed model was implemented on the RAVDESS and IEMOCAP data corpora and compared with the CNN-SVM and CNN-RF models for SER. The TDACNN-RF model exhibited test classification accuracies of 92.19% and 90.27% on the RAVDESS and IEMOCAP data corpora, respectively. The experimental results proved that the proposed model is efficient in extracting spatiotemporal features from time-series speech signals and can classify emotions with good accuracy. The class confusion among the emotions was reduced for both data corpora, proving that the model achieved generalization.

Keywords: Speech Emotion Recognition, Human-Computer Interaction, Ensemble classifiers, Random Forest, Time distributed Layers, Attention Layers, Convolution Neural Network, Spatiotemporal features.

 

 
CONCENTRATION SEPARATION PREDICTION MODEL TO ENHANCE PREDICTION ACCURACY OF PARTICULATE MATTER
1Yong-jin Jung & 1Chang-heon Oh
1Department of Electrical, Electronics, and Communication Engineering, Korea University of Technology and Education(KOREATECH), Korea
yj0211@koreatech.ac.kr; *choh@koreatech.ac.kr
 
 
ABSTRACT
 

Demand for more accurate particulate matter forecasts is accumulating owing to increased interest and issues regarding the particulate matter. Incredibly low-concentration particulate matter, which accounts for most of the overall particulate matter, is often underestimated when a particulate matter prediction model based on machine learning is used. Thus, we proposed a concentration-specific separation prediction model to overcome this shortcoming. Three prediction models based on deep neural network (DNN), recurrent neural network (RNN), and long short-term memory (LSTM), commonly used for performance evaluation of the proposed prediction model, are used as comparative models. Root mean square error (RMSE), mean absolute percentage error (MAPE), and accuracy was used for performance evaluation, and the results show that the prediction accuracy for all air quality index (AQI) segments were more than 80% in the entire concentration spectrum. In addition, we confirmed that the overprediction phenomenon of single neural network models concentrated in the 'normal' AQI region was alleviated.

Keywords: SDNN, RNN, LSTM, Particulate Matter.

 

 
A NOVEL METHOD FOR FASHION CLOTHING IMAGE CLASSIFICATION BASED ON DEEP LEARNING
1Seong-Yoon Shin, 2Gwanghyun Jo & 3Guangxing Wang
1School of Software, Kunsan National University, South Korea
2Department Mathematics, Kunsan National University, South Korea
3School of Computer and Big Data Science, Jiujiang University, China
s3397220@kunsan.ac.kr; *gwanghyun@kunsan.ac.kr; wanggxrs@163.com
*Corresponding author
 
 
ABSTRACT
 

Image recognition and classification is a significant research topic in computational vision and a widely used computer technology.  The methods often used in image classification and recognition tasks are based on deep learning, like Convolutional Neural Networks (CNN), LeNet, and Long Short-Term Memory networks (LSTM). Unfortunately, the classification accuracy of these methods is not satisfactory. In recent years, using large-scale deep learning networks to achieve image recognition and classification can improve classification accuracy, such as VGG16 and ResNet. However, due to the deep network hierarchy and complex parameter settings, these models take more time in the training phase, especially when the sample number is small, which can easily lead to over-fitting. This paper suggests a deep learning-based image classification technique based on a CNN model, and improved convolutional and pooling layers.  Furthermore, we adopt the approximate dynamic learning rate update algorithm in the model training to realize the learning rate's self-adaptation, ensure the model's rapid convergence, and shorten the training time. Using the proposed model, we conducted an experiment on the Fashion-MNIST dataset, taking 6000 images as the training data set and 1000 images as the test data set.  In actual experiments, the classification accuracy of the suggested method was 93%, 4.6% higher than that of the basic CNN model.  Simultaneously, we compared the influence of the batch size of model training on classification accuracy. Experimental outcomes show our way is very generalized in fashion clothing image classification tasks.

Keywords: Machine Learning, Deep Learning, Computational Vision, Convolution Neural Network(CNN), Fashion Clothing Image Classification

 

 
INVESTIGATING JOB MISMATCH IN SOFTWARE INDUSTRY THROUGH NEWS BIG DATA
1Juho Song, 2Ho Lee & 3Oh-Young Kwon
1Policy Statistics Team, Korea Software Industry Association, Korea
2,3Department of Future Technology, Korea University of Technology & Education, Korea
juho@sw.or.kr; *leeho32@koreatech.ac.kr; oykwon@koreatech.ac.kr
*Corresponding Author
 
 
ABSTRACT
 

The purpose of this study is to identify issues related to software manpower, which became more important in the era of the 4th Industrial Revolution in Korea. Through this, the results of this study can provide guidelines for those who establish software manpower training policies for solving the software industry's human resource paradox. As a research method, we used quantitative text network analysis and qualitative analysis from industry experts to interpret the results. A total of 14,752 news data mentioning software manpower were extracted, and data pre-processing for the synonyms and negative words were performed. The network is non-directional and consists of 14,074 words (nodes) and 1,542,383 word combinations (edges). In addition, the network was clustered based on Modularity, and the degree of connection and eigenvector centrality were used to determine the importance of nodes. Analysis of the results showed that the government's efforts through the Korean Ministry of Science and ICT were vital in creating jobs that fueled software innovation growth, and that software education was actively promoted in order to develop software talent. This study has the following implications. It was confirmed that software is making a high contribution to the expansion of business opportunities and job creation in the fields of new technology and software convergence technology. To resolve the software manpower supply-demand mismatch, it is necessary to cultivate high-quality software talent and provide mid- to long-term activities in order to attract competent human resources. In addition, in order to strengthen national software competitiveness, it is necessary to develop and expand programs that link education and recruitment in terms of public-private cooperation along with government-led investment.

Keywords: News big data, Network analysis, Software manpower, Human resource development.

 

 
PREDICTION OF STUDENTS’ SOCIAL WELL-BEING BASED ON PERSONALITY TRAITS DETERMINANTS
*1Nur Atiqah Rochin Demong, 2Melissa Shahrom, 3Ramita Abdul Rahim, 4Emi Normalina Omar & 5Mornizan Yahya
12345Department of Technology and Supply Chain Management Studies, Faculty of Business and Management, Universiti Teknologi MARA, Malaysia
rochin@uitm.edu.my; melissa@uitm.edu.my; ramita@uitm.edu.my; emi128@uitm.edu.my; mornizan@uitm.edu.my
 
 
ABSTRACT
 

Social well-being is a field that analyses society, individual behavioural patterns, behavioural networks, and cultural elements of daily life. Social well-being develops critical thinking by understanding the social frameworks that affect humans by exposing the social basis of daily actions. For instance, when students are pleased, their academic achievement, behaviour, social integration, and happiness improve. This study predicts the effects of the Big 5 Personality Traits (Extraversion, Openness, Agreeableness, Emotional Stability, and Conscientiousness) on students’ Industry 4.0 Social Well-being level by analysing their demographic and personality traits. A dataset was gathered through a survey distributed to students in a selected institution. The accuracy of the classifier was evaluated using the WEKA tool on a dataset of 286 instances and 19 attributes, for which a confusion matrix was generated. After evaluating the results of all algorithms, it was discovered that IBk and Randomizable Filtered Classifier algorithms provide the highest accuracy with a similar percentage value of 91.26% on the social well-being readiness. The agreeableness personality trait, which represents a person’s level of pleasantness, politeness, and helpfulness, had the greatest influence on the social well-being of the students. They have a positive outlook on human behaviour and get along well with others. Since social well-being contributes to a person’s increased quality of life and happiness, improving students’ current quality of life would lead to the development of a social parameter that can assess the growth of a country and the increased happiness of families and communities.

Keywords: WEKA, Personality Traits, Social Well-being, Data Mining, Classification Algorithms