Kok-Chin Khor
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Kok-Chin Khor has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Keng-Hoong Ng
Universiti Tunku Abdul Rahman, Sungai Long Campus, Jalan Sungai Long, Bandar Sungai Long
13 shared publications - Somnuk Phon-Amnuaisuk
Universiti Teknologi Brunei
7 shared publications Choo-Yee Ting
Multimedia University
6 shared publications- Choo-Yee Ting
Centre for Big Data and Blockchain Technologies, CoE of Advanced Cloud, Multimedia University, Cyberjaya 63100, Malaysia
4 shared publications Gee-Kok Tong
Multimedia University, Persiaran Multimedia, Cyberjaya, 63100, Malaysia
4 shared publications- Meei-Hao Hoo
Universiti Tunku Abdul Rahman
4 shared publications
Latest Publications
- Open Access
- 13 August 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (8) , e0329146
Cross-attention guided discriminative feature selection for robust point cloud domain generalization
In recent years, deep learning networks have been widely employed for point cloud classification. However, discrepancies between training and testing scenarios often result in erroneous predictions. Domain generalization (DG) aims to achieve high classification accuracy in unseen scenarios without requiring additional training. Although current DG methodologies effectively employ data augmentation and representation learning, they inadvertently neglect a key component: discriminative feature selection, which we identify as a crucial missing element for achieving robust domain generalization. To fully leverage the geometric features of point clouds, we propose a novel domain generalization method that emphasizes transferring contextual information to improve generalization performance for 3D point clouds. Our method projects the point cloud into multiple views and employs a 2D adaptive feature extractor to capture and aggregate weighted semantic features, while leveraging the DGCNN network to extract 3D spatial geometric features. Additionally, we incorporate an attention mechanism to fuse 2D semantic features with 3D geometric features, facilitating the selection of discriminative features from point clouds. The experiments demonstrate that our method outperforms state-of-the-art methods in both multi-source and single-source tasks, achieving superior generalization performance.
- 13 August 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (8) , e0329146
Open AccessCross-attention guided discriminative feature selection for robust point cloud domain generalization
In recent years, deep learning networks have been widely employed for point cloud classification. However, discrepancies between training and testing scenarios often result in erroneous predictions. Domain generalization (DG) aims to achieve high classification accuracy in unseen scenarios without requiring additional training. Although current DG methodologies effectively employ data augmentation and representation learning, they inadvertently neglect a key component: discriminative feature selection, which we identify as a crucial missing element for achieving robust domain generalization. To fully leverage the geometric features of point clouds, we propose a novel domain generalization method that emphasizes transferring contextual information to improve generalization performance for 3D point clouds. Our method projects the point cloud into multiple views and employs a 2D adaptive feature extractor to capture and aggregate weighted semantic features, while leveraging the DGCNN network to extract 3D spatial geometric features. Additionally, we incorporate an attention mechanism to fuse 2D semantic features with 3D geometric features, facilitating the selection of discriminative features from point clouds. The experiments demonstrate that our method outperforms state-of-the-art methods in both multi-source and single-source tasks, achieving superior generalization performance.
- Open Access
- 17 September 2024
- Published by Springer Nature
- p. C1
Correction to: Predicting Undergraduate Academic Success with Machine Learning Approaches
- 17 September 2024
- Published by Springer Nature
- p. C1
Open AccessCorrection to: Predicting Undergraduate Academic Success with Machine Learning Approaches
- 30 July 2024
- Published by Springer Nature
- p. 144-153
Predicting Undergraduate Academic Success with Machine Learning Approaches
- 30 July 2024
- Published by Springer Nature
- p. 144-153
Predicting Undergraduate Academic Success with Machine Learning Approaches
- 30 July 2024
- Published by Springer Nature
- p. 43-52
Evaluating Deep Transfer Learning Models for Detecting Various Face Mask Wearings
- 30 July 2024
- Published by Springer Nature
- p. 43-52
Evaluating Deep Transfer Learning Models for Detecting Various Face Mask Wearings
- 30 July 2024
- Published by Springer Nature
- p. 165-174
Evaluating Path-Finding Algorithms for Real-Time Route Recommendation System Built using FreeRTOS
- 30 July 2024
- Published by Springer Nature
- p. 165-174
Evaluating Path-Finding Algorithms for Real-Time Route Recommendation System Built using FreeRTOS
- Open Access
- 30 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 77486-77500
PointADAM: Unsupervised Adversarial Domain Adaptation on Point Clouds With Metric Learning via Compact Feature Representation
Domain adaptation can mitigate the problem of limited labels in deep learning training. Nevertheless, extending the 2D domain adaptation method directly to 3D encounters challenges unique to point clouds, frequently resulting in inadequate feature alignment and a lack of discriminative features for decision boundaries. In light of this, we propose an unsupervised adversarial domain adaptation with metric learning (PointADAM) via compact feature representation. PointADAM is a two-stage architecture. In the first stage, it extracts high-level semantic and low-level local features from source and target domains based on the geometric relations of point cloud neighbors. Subsequently, it aligns the features from the two domains in an unsupervised adversarial manner, thereby obtaining a shared common feature space. In the second stage, a self-training method is employed to label the target domain with pseudo-labels. It is followed by selecting target domain samples with high confidence. To enhance discriminative feature learning, joint supervision is employed through both cross-entropy loss and center loss. Cross-entropy is utilized to calculate the probability of each category for classification and center loss is utilized to compact intra-class distribution and disperse inter-class distance, thereby improving classification accuracy by optimizing the decision boundary. Experimental results on the PointDA-10 dataset illustrate that our method performs relatively better than the state-of-the-art methods for domain adaptation on the point cloud by 3.1% in accuracy.
- 30 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 77486-77500
Open AccessPointADAM: Unsupervised Adversarial Domain Adaptation on Point Clouds With Metric Learning via Compact Feature Representation
Domain adaptation can mitigate the problem of limited labels in deep learning training. Nevertheless, extending the 2D domain adaptation method directly to 3D encounters challenges unique to point clouds, frequently resulting in inadequate feature alignment and a lack of discriminative features for decision boundaries. In light of this, we propose an unsupervised adversarial domain adaptation with metric learning (PointADAM) via compact feature representation. PointADAM is a two-stage architecture. In the first stage, it extracts high-level semantic and low-level local features from source and target domains based on the geometric relations of point cloud neighbors. Subsequently, it aligns the features from the two domains in an unsupervised adversarial manner, thereby obtaining a shared common feature space. In the second stage, a self-training method is employed to label the target domain with pseudo-labels. It is followed by selecting target domain samples with high confidence. To enhance discriminative feature learning, joint supervision is employed through both cross-entropy loss and center loss. Cross-entropy is utilized to calculate the probability of each category for classification and center loss is utilized to compact intra-class distribution and disperse inter-class distance, thereby improving classification accuracy by optimizing the decision boundary. Experimental results on the PointDA-10 dataset illustrate that our method performs relatively better than the state-of-the-art methods for domain adaptation on the point cloud by 3.1% in accuracy.
- Open Access
- 31 March 2024
- Published by Society of Visual Informatics in JOIV : International Journal on Informatics Visualization
- Vol. 8 (1) , 115
Clustering Defensive Shariah-compliant Stocks Using Financial Performance as the Indicator
Malaysian stocks, including Shariah-compliant stocks, have experienced turbulence last year. Although there are defensive stocks, the well-performing ones are not easily identified. Researchers have proposed various metrics to identify defensive stocks. However, most of the approaches require human intervention. In this study, we focus on Shariah-compliant stocks and propose to automate the labeling of stocks in terms of their financial performance via clustering. The study aims to identify the optimal clustering method to label the clusters. This was achieved by first employing k-Means, Agglomerative, and Mean Shift clustering to group similar stocks before labeling. When labeling, the criteria to distinguish well-performing defensive Shariah-compliant stocks were high dividend yield, low price-earnings ratio, low Beta value, and low price-to-book value. After labelling each stock with its financial performance (Low, Medium, High), we performed classification using Logistic Regression, k-Nearest Neighbors, Support Vector Machine, Decision Tree, and Random Forest to verify the credibility of the labels. Based on the results, the clusters created by k-Means clustering outperformed the rest in matching accuracy. Further investigation was conducted on the k-Means data set by dividing the data according to sector and classifying each sector’s data separately. Logistic Regression outperformed other classification algorithms with an accuracy of 71.5%. The findings also suggested accuracy increased when stocks were classified according to sectors. Further considerations include performing outlier analysis on the data to select well-performing stocks.
- 31 March 2024
- Published by Society of Visual Informatics in JOIV : International Journal on Informatics Visualization
- Vol. 8 (1) , 115
Open AccessClustering Defensive Shariah-compliant Stocks Using Financial Performance as the Indicator
Malaysian stocks, including Shariah-compliant stocks, have experienced turbulence last year. Although there are defensive stocks, the well-performing ones are not easily identified. Researchers have proposed various metrics to identify defensive stocks. However, most of the approaches require human intervention. In this study, we focus on Shariah-compliant stocks and propose to automate the labeling of stocks in terms of their financial performance via clustering. The study aims to identify the optimal clustering method to label the clusters. This was achieved by first employing k-Means, Agglomerative, and Mean Shift clustering to group similar stocks before labeling. When labeling, the criteria to distinguish well-performing defensive Shariah-compliant stocks were high dividend yield, low price-earnings ratio, low Beta value, and low price-to-book value. After labelling each stock with its financial performance (Low, Medium, High), we performed classification using Logistic Regression, k-Nearest Neighbors, Support Vector Machine, Decision Tree, and Random Forest to verify the credibility of the labels. Based on the results, the clusters created by k-Means clustering outperformed the rest in matching accuracy. Further investigation was conducted on the k-Means data set by dividing the data according to sector and classifying each sector’s data separately. Logistic Regression outperformed other classification algorithms with an accuracy of 71.5%. The findings also suggested accuracy increased when stocks were classified according to sectors. Further considerations include performing outlier analysis on the data to select well-performing stocks.
- Open Access
- 6 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 23636-23652
Detecting Frauds and Payment Defaults on Credit Card Data Inherited With Imbalanced Class Distribution and Overlapping Class Problems: A Systematic Review
Credit card payments are one popular e-payment option apart from cash payments. Recent reports show that credit card fraud and payment defaults are increasing annually and are alarming. Thus, researchers have attempted various machine learning techniques to address these two challenges. However, they are challenged to mitigate the two major problems inherited in credit card data: (i) imbalanced class distribution and (ii) overlapping classes. Mitigating these problems shall effectively detect credit card frauds and payment defaults, thus benefiting card issuers and holders. Hence, this paper aims to develop a systematic review using PRISMA to identify and compare various credit card datasets, machine learning techniques, and evaluation metrics. Subsequently, we provide recommendations for handling these two problems. We extracted research papers from 2016 to 2023 from ScienceDirect, Springer, Association and Computing Machinery (ACM), and IEEE databases. The papers shall be included if written in English and published in peer-reviewed and indexed journals or conference proceedings. Finally, 87 papers were selected based on the eligibility criteria. Based on our findings, the European and Taiwan datasets are widely used in the research community. However, most researchers focus on tackling imbalanced class distribution rather than two problems together. We recommended to the research community the application of deep learning, ensemble learning, and sampling methods to effectively detect fraud and payment defaults on credit card datasets that inherit the two problems. In evaluating the machine learning algorithms, we recommend using metrics that can separately evaluate the algorithms’ performance in detecting frauds/payment defaults and normal transactions.
- 6 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 23636-23652
Open AccessDetecting Frauds and Payment Defaults on Credit Card Data Inherited With Imbalanced Class Distribution and Overlapping Class Problems: A Systematic Review
Credit card payments are one popular e-payment option apart from cash payments. Recent reports show that credit card fraud and payment defaults are increasing annually and are alarming. Thus, researchers have attempted various machine learning techniques to address these two challenges. However, they are challenged to mitigate the two major problems inherited in credit card data: (i) imbalanced class distribution and (ii) overlapping classes. Mitigating these problems shall effectively detect credit card frauds and payment defaults, thus benefiting card issuers and holders. Hence, this paper aims to develop a systematic review using PRISMA to identify and compare various credit card datasets, machine learning techniques, and evaluation metrics. Subsequently, we provide recommendations for handling these two problems. We extracted research papers from 2016 to 2023 from ScienceDirect, Springer, Association and Computing Machinery (ACM), and IEEE databases. The papers shall be included if written in English and published in peer-reviewed and indexed journals or conference proceedings. Finally, 87 papers were selected based on the eligibility criteria. Based on our findings, the European and Taiwan datasets are widely used in the research community. However, most researchers focus on tackling imbalanced class distribution rather than two problems together. We recommended to the research community the application of deep learning, ensemble learning, and sampling methods to effectively detect fraud and payment defaults on credit card datasets that inherit the two problems. In evaluating the machine learning algorithms, we recommend using metrics that can separately evaluate the algorithms’ performance in detecting frauds/payment defaults and normal transactions.
- Open Access
- 28 September 2023
- Published by Hong Kong Success Culture Press in Journal of System and Management Sciences
- Vol. 13 (5)
Portfolio Construction of Good Defensive Malaysian Shariah-Compliant Stocks Using Data Mining Techniques
- 28 September 2023
- Published by Hong Kong Success Culture Press in Journal of System and Management Sciences
- Vol. 13 (5)
Open AccessPortfolio Construction of Good Defensive Malaysian Shariah-Compliant Stocks Using Data Mining Techniques
- 22 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Games
- Vol. 16 (3) , 577-585
Profiling and Identifying Smurfs or Boosters on Dota 2 Using K-Means and IQR
Dota 2 is one popular multiplayer online battle arena game, and it holds the grandest e-sports tournament in the world—The International. However, smurfs and boosters are plaguing the game, causing a continuous decline in the player count. Smurfs are skilled players who stomp less experienced players, while boosters are paid to improve players' rank. At this stage, the developers have brought updates on smurf detection based on players' complaints, where smurf accounts are likely to be prevented from entering the game. This article proposes a smurf or booster detection among the players by profiling and identifying them based on statistical differences in features. Initially, we created a dataset with player data collected from the OpenDota API. Then, K-means was used to group and profile the players. Subsequently, the interquartile range method was applied to the high-performing players to identify the smurfs or boosters. We then invited three Dota 2 game experts to review the resulting profiles. A 95% accuracy score was achieved using majority voting. The methodology proposed in this article can be implemented in the Dota 2 to detect smurfs or boosters automatically. The findings in this article shall contribute to prolonging the game's life span.
- 22 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Games
- Vol. 16 (3) , 577-585
Profiling and Identifying Smurfs or Boosters on Dota 2 Using K-Means and IQR
Dota 2 is one popular multiplayer online battle arena game, and it holds the grandest e-sports tournament in the world—The International. However, smurfs and boosters are plaguing the game, causing a continuous decline in the player count. Smurfs are skilled players who stomp less experienced players, while boosters are paid to improve players' rank. At this stage, the developers have brought updates on smurf detection based on players' complaints, where smurf accounts are likely to be prevented from entering the game. This article proposes a smurf or booster detection among the players by profiling and identifying them based on statistical differences in features. Initially, we created a dataset with player data collected from the OpenDota API. Then, K-means was used to group and profile the players. Subsequently, the interquartile range method was applied to the high-performing players to identify the smurfs or boosters. We then invited three Dota 2 game experts to review the resulting profiles. A 95% accuracy score was achieved using majority voting. The methodology proposed in this article can be implemented in the Dota 2 to detect smurfs or boosters automatically. The findings in this article shall contribute to prolonging the game's life span.