WKWee How Khoh

Wee How Khoh

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Wee How Khoh has a total of 45 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Contextual Contrastive Representation Learning with LightGBM for Intrusion Detection in Heterogeneous IoT Networks

    by Sunil Kumar, Sameeksha Khare, Isha Chopra, Jugul Kishor Gupta, Devendra Kumar, Wee How Khoh, Firoz Khan, Vivek Kumar Sharma

    The increasing adoption of Internet of Things (IoT) technologies in smart farming has improved agricultural automation and decision-making while expanding the attack surface of interconnected devices and communication networks. Intrusion detection in these environments is challenged by heterogeneous network features, severe class imbalance, and evolving attack patterns. This study proposes CCRL-LightGBM, a contextual contrastive representation learning framework that integrates TabTransformer-based feature learning, supervised contrastive refinement, and Light Gradient Boosting Machine (LightGBM) classification. The framework was evaluated on the CICIoT2023 and CIC IoT-DIAD 2024 datasets using a unified preprocessing pipeline and compared with representative machine learning and deep learning models. Under random-stratified evaluation, CCRL-LightGBM achieved an accuracy of 99.63% and F1-score of 99.81% on CICIoT2023 and an accuracy of 99.39% and F1-score of 99.68% on CIC IoT-DIAD 2024. GPU inference latency was 3.78 ms per sample under the evaluated experimental conditions. In paired comparisons with TT–LightGBM, the CCRL-LightGBM configuration achieved higher F1-scores under temporal, source-disjoint, and device-disjoint evaluation protocols, with device-disjoint F1-scores of 93.31% on CICIoT2023 and 92.56% on CIC IoT-DIAD 2024. However, performance declined under stricter partitioning protocols, highlighting the challenges of generalization to temporally shifted traffic and previously unseen source or device groups. Unadapted cross-dataset evaluation further revealed limitations in transfer performance between the two benchmark environments. These results demonstrate strong benchmark-level performance and improved results relative to the baseline under the evaluated conditions while highlighting the need for further validation using agriculture-specific traffic and deployment-level measurements before operational suitability for smart-farming environments can be established.

  • Open Access

    Inherently Interpretable Kawasaki Disease Prediction Using Concept Bottleneck Models, Graph-Structured Concept Interactions, and Counterfactual Clinical Reasoning

    by Tina Babu, Rekha R. Nair, Wee How Khoh, Jashila Nair Mogan, Sindhu S.

    Accurate differential diagnosis of Kawasaki disease (KD) from other febrile conditions remains challenging owing to nonspecific clinical presentation, absence of validated biomarkers, and reliance on post hoc explainability methods that cannot support clinician-level counterfactual reasoning. We propose ClinCBM-KD, an inherently interpretable concept bottleneck model routing all predictions through a clinically predefined 21-concept bottleneck aligned with AHA diagnostic criteria; the architecture integrates sparse concept encoders, a learnable concept interaction graph (CIG) modelling pairwise concept dependencies, and a concept-space gradient-boosting predictor, jointly optimised to enforce concept supervision, feature sparsity, and clinical graph priors. Evaluated on the Pediatric Heart Network registry (n=645; 77 analytical variables) via a stratified 80/20 training, held-out test split with ten-fold cross-validation within the training partition, ClinCBM-KD achieved an AUC-ROC of 0.908, accuracy of 87.3%, recall of 91.8%, and F1-score of 90.0%, with no statistically significant AUC difference versus the strongest post hoc baseline (logistic regression, AUC 0.910; DeLong p=0.741), concept completeness Γ=0.921, soundness Ω=0.884, and the lowest Brier score (0.094) of all classifiers; ablation confirmed concept supervision as the dominant contributor (ΔAUC=0.040). A temporally restricted evaluation retaining only c1–c17 (excluding treatment-derived and echocardiographic concepts) confirmed sustained diagnostic performance. Within this single-cohort retrospective study, ClinCBM-KD achieves diagnostic performance equivalent to unconstrained post hoc models while enabling transparent, concept-level predictions and counterfactual clinical reasoning; prospective human-centred validation is required to establish whether these capabilities translate to improved clinical practice.

  • Open Access

    AI-Driven Intrusion Detection System for Network Security

    by Wee How Khoh, Zhi Lin Sarah Teoh, Hui Yen Yap, Pin Shen Teh

    In the evolving landscape of cybersecurity, Intrusion Detection Systems (IDS) play a vital role in safeguarding computer networks against malicious activity. Traditional signature-based IDS approaches are increasingly ineffective in detecting novel, complex, or zero-day attacks due to their reliance on predefined rules. To overcome these limitations, this study proposes an AI-driven IDS that integrates both classical machine learning and modern deep learning techniques. The framework introduces and compares three models such as Support Vector Machine (SVM), Convolutional Neural Network (CNN), and a hybrid CNN-SVM model. The system is designed to analyze network traffic patterns and classify them as either benign or malicious, enhancing detection capabilities through intelligent feature learning and classification. Two widely recognized benchmark datasets are used to train and validate the models, ensuring the system's applicability to a variety of network environments. Furthermore, the research includes the development of a real-time detection component that incorporates live packet capture, feature extraction, and dynamic visualization via a dashboard interface. This paper contributes to the field by demonstrating how hybrid AI models can effectively address the challenges of network intrusion detection. The study emphasizes the importance of combining traditional and deep learning approaches to build scalable, adaptive, and accurate intrusion detection systems for modern network infrastructures.

  • Open Access

    An Interpretable and Uncertainty-Aware Deep Learning Framework for Early Sepsis Prediction Using SHAP-Enhanced Attention and Continuous-Time Neural Networks

    by Rekha R. Nair, Tina Babu, Balamurugan Balusamy, Wee How Khoh, Alaa M. Momani, Basem Abu Zneid

    Sepsis is a prominent cause of death in intensive care units, and delayed diagnosis greatly worsens fatal outcomes due to the complex, irregular, and uneven character of clinical time-series data. Hence we proposed an interpretable and uncertainty-aware deep learning architecture that solves data quality, temporal irregularity, and clinical explainability restrictions, which are frequently addressed separately by existing models. The suggested method combines Bidirectional Recurrent Imputation for Time Series (BRITS)-based imputation, hybrid Conditional Tabular Generative Adversarial Network-Synthetic Minority Over-sampling Technique (CTGAN-SMOTE) data augmentation, a Temporal Convolutional Networks (TCN)-Attention architecture, and continuous-time neural Ordinary Differential Equations (ODEs), along with SHapley Additive exPlanations (SHAP)-based feature attribution and uncertainty quantification. The experimental evaluation on a large ICU dataset reveals greater predictive accuracy, with an AUROC of 0.926 and accurate early warnings up to six hours before clinical onset, all while maintaining strong interpretability and calibration. The proposed framework demonstrates strong predictive performance and provides early warnings up to six hours before clinical onset, while maintaining interpretability and calibration. While the results are promising, further validation across multiple clinical settings is required to confirm its generalisability and real-world applicability.

  • Open Access

    AI Creation of Facial Expression Database for Advanced Emotion Recognition Using Diffusion Model and Pre-Trained CNN Models

    by Jia Jun Ho, Wee How Khoh, Ying Han Pang, Hui Yen Yap, Fang Chuen Lim Alvin

    With applications in psychology, security, and human–computer interaction, facial expression recognition (FER) has become an essential tool for non-verbal communication. Current research often categorizes expressions into micro- and macro-types, yet existing datasets suffer from inconsistent labelling for classes, limited diversity of the databases, and insufficient scale for the currently available datasets. To address these gaps, this work proposes a novel framework combining the diffusion model with pre-trained CNNs. Leveraging original images from established datasets, CASME II, we generate synthetic facial expressions to augment training data, mitigating bias and inconsistency. The synthetic dataset is evaluated using ResNet 50, VGG16 and Inception V3 architectures. Inception V3 trained on the proposed AI-generated dataset and tested using CASME II, VGG-16 with data augmentation applied is trained on CASME II and tested on the proposed AI-generated dataset, and Inception V3 with 30% freezing layers method is trained on the proposed AI-generated dataset and tested using CASME II. These all successfully achieved state-of-the-art performance. The data augmentation and freezing layers approaches significantly improved the performance of the models. Our proposed approaches achieved state-of-the-art performance and outperformed most of the existing state-of-the-art approaches benchmarked in this study.

  • Open Access

    FDSTCN-EEG: Federated Depthwise Separable Temporal Convolutional Networks for Decentralized EEG Seizure Detection

    by Zheng You Lim, Ying Han Pang, Shih Yin Ooi, Wee How Khoh, Yee Jian Chew

    In this paper, we propose FDSTCN-EEG, which is a customized federated learning framework for EEG-based seizure detection that leverages deep depthwise separable temporal convolutions and asynchronous model aggregation. The network design tackles major problems in distributed healthcare AI by jointly boosting computational efficiency, training rate, and classification performance. In this paper, we propose FDSTCN-EEG, a novel federated learning framework specifically designed for EEG-based seizure detection. Our key contributions are threefold: first, high architectural efficiency with depthwise separable temporal convolutions, reducing parameters by 40.4% (9.8M to 5.8M) while maintaining accuracy of 96.96%; second, speeding up training by a factor of 38.5% compared with synchronous learning via an asynchronous aggregation protocol; finally, a privacy-preserving decentralized learning model without sharing raw EEG data and with the capability of coping with the heterogeneous clinical technology infrastructure. Extensive experiments show superior performance (accuracy: 96.96%, F1-score: 97.02%) and practical viability for real-world seizure monitoring systems. Such work introduces a practical privacy-preserving medical AI paradigm, which balances model efficiency with training scalability and clinical quality accuracy.

  • Open Access

    AI Creation of Facial Expression Database for Advanced Emotion Recognition

    by Jia Jun Ho, Wee How Khoh, Ying Han Pang, Hui Yen Yap, Fang Chuen Lim Alvin

    With applications in psychology, security, and human-computer interaction, facial expression recognition (FER) has become an essential tool for non-verbal communication. Current research often categorizes expressions into micro and macro types, yet existing datasets suffer inconsistent labelling for classes, limited diversity of the databases, and insufficient scale for the currently available datasets. To address these gaps, this work proposes a novel framework combining the Diffusion model with pre-trained CNNs. Leveraging original images from established datasets, CASME â…¡, we generate synthetic facial expressions to augment training data, mitigating bias and inconsistency. The synthetic dataset is evaluated using ResNet 50, VGG16 and Inception V3 architectures. Inception V3 trained on the proposed AI-generated dataset and tested using CASME â…¡ achieved the highest accuracy of 99.48%. VGG-16 with data augmentation applied was trained on CASME â…¡ and tested on the proposed AI-generated dataset achieved 99.54%. While 30% freezing layers method is utilized, Inception V3 trained on the proposed AI-generated dataset and tested using CASME â…¡ obtained an accuracy of 99.53%. The data augmentation and freezing layers approaches have significantly improved the performance of the models. Our proposed approaches have achieved state-of-the-art performance and outperformed most of the existing state-of-the-art approaches benchmarked in this study.

  • Open Access

    AgriFewNet—A Lightweight RGB Few-Shot Learning Model for Efficient Plant Disease Classification

    by Tina Babu, Rekha R. Nair, Balamurugan Balusamy, Wee How Khoh, Jashila Nair

    The rapid growth of artificial intelligence (AI) has enabled efficient crop disease detection even in data-scarce agricultural settings. This study proposes AgriFewNet, a few-shot learning framework designed to improve classification accuracy using RGB imagery captured from publicly available datasets. The objective is to enable fast model adaptation to new disease classes using minimal labeled samples while maintaining high reliability in real-world conditions. AgriFewNet employs a hierarchical attention-enhanced ResNet-18 backbone incorporating dual spatial and channel attention to extract discriminative RGB features. A Model-Agnostic Meta-Learning (MAML) approach facilitates quick generalization to previously unexplored illness categories, while a prototype-based classifier guarantees compact representation learning. Using only RGB images, experiments on the PlantVillage and New PlantVillage datasets produced accuracies of 87.3% (1-shot), 94.8% (5-shot), and 97.1% (10-shot), surpassing leading few-shot baselines by as much as 7.9%. The findings show that AgriFewNet offers a resource-efficient and scalable method for intelligent crop monitoring, enhancing food security and precision agriculture.


  • Open Access

    NeuroFed-LightTCN: Federated Lightweight Temporal Convolutional Networks for Privacy-Preserving Seizure Detection in EEG Data

    by Zheng You Lim, Ying Han Pang, Shih Yin Ooi, Wee How Khoh, Yee Jian Chew

    This study investigates on-edge seizure detection that aims to resolve two major constraints that hold the deployment of deep learning models in clinical settings at present. First, centralized training requires gathering and consolidating data across institutions, which poses a serious issue of privacy. Second, a high computational overhead inherent in inference imposes a crushing burden on resource-limited edge devices. Hence, we propose NeuroFed-LightTCN, a federated learning (FL) framework, incorporating a lightweight temporal convolutional network (TCN), designed for resource-efficient and privacy-preserving seizure detection. The proposed framework integrates depthwise separable convolutions, grouped with structured pruning to enhance efficiency, scalability, and performance. Furthermore, asynchronous aggregation is employed to mitigate training overhead. Empirical tests demonstrate that the network can be reduced fully to 70% with a 44.9% decrease in parameters (65.4 M down to 34.9 M and an inferencing latency of 56 ms) and still maintain 97.11% accuracy, a metric that outperforms both the non-FL and FL TCN optimizations. Ablation shows that asynchronous aggregation reduces training times by 3.6 to 18%, and pruning sustains performance even at extreme sparsity: an F1-score of 97.17% at a 70% pruning rate. Overall, the proposed NeuroFed-LightTCN addresses the trade-off between computational efficiency and model performance, delivering a viable solution to federated edge-device learning. Through the interaction of federated-optimization-driven approaches and lightweight architectural innovation, scalable and privacy-aware machine learning can be a practical reality, without compromising accuracy, and so its potential utility can be expanded to the real world.


  • Multi-head Temporal Learning for Human Activity Recognition Using Smartphone

    by So Zheng Wei, Pang Ying Han, Khoh Wee How, Ooi Shih Yin