Institutional-Repository, University of Moratuwa
Welcome to the University of Moratuwa Digital Repository, which houses postgraduate theses and dissertations, research articles presented at conferences by faculties and departments, university-published journal articles and research publications authored by academic staff. This online repository stores, preserves and distributes the University's scholarly work. This service allows University members to share their research with a larger audience.
 Research Publications |  Thesis & Dissertation |  E- Books |
 UoM Journal Publications |  UoM Conference Proceedings |  Articles published in Scimago's Q1 journals |  UoM Research Reports |  Other Articles authored by UoM staff |
Recent Submissions
item: Conference-Extended-Abstract
Smart mirror application for social event based outfit recommendation and skin health analysis
(Department of Computer Science and Engineering, 2025) Rathnasekara, N; Damith, N; Herath, D; Seneviratne, O; Wijenayake, U; Athuraliya, CD
The rapid evolution of Artificial Intelligence(AI) has paved the way for innovative applications in everyday objects. Among these, a smart mirror is a representation of a traditional mirror with integrated technologies to provide realtime personalized recommendations and health insights. Busy schedules make it challenging for individuals to manage their to-do lists, health, and even dress code choices. According to the past work done, both Khandaker et al. [1] and Simone et al. [2] developed smart mirrors with various features, such as face verification, weather updates, calendar events, news, traffic, and emotion recognition. Considering social event-based fashion recommendation, Federico et al [3] proposed an event classifier combined with a recommender by implementing a fashion object detector for social events. Nikita et al [4] proposed outfit recommendation by detecting user outfits, identifying the social event, and suggesting similar outfits for the detected event. However the system lacks the ability of considering outfits suitable for multiple social events. In the context of acne detection, Quan et al. [5] proposed an automatic system that employs Faster Region-based Convolutional Neural Network (R-CNN) for acne lesion detection and LightGBM for severity grading. Similarly, Hang et al. [6] introduced an ensemble neural network-based approach featuring a classification module for severity estimation and an acne localization module. However, prior research has focused on general acne detection, there is limited work addressing localizing acne within the ”Danger Triangle of the Face” which is a crucial clinical aspect. This research proposes a comprehensive Internet of Things(IoT) smart mirror framework to address past limitations. Mainly, it has features for face verification, acne detection with ”Triangle of Death” localization, and social event-based fashion recommendation. By leveraging these components, the proposed smart mirror represents a significant advancement towards smart computing, seamlessly integrated with digital intelligence.
item: Conference-Extended-Abstract
Developing a question answering system for the Sri Lankan school education system
(Department of Computer Science and Engineering, 2025) Kiridana, YMWHMRPJRB; Gihan Dias, G; Athuraliya, CD
The integration of Artificial Intelligence (AI) into education unlocks opportunities for personalized learning. However, low-resource languages such as Sinhala currently lack robust Natural Language Processing (NLP) tools. This paper proposes a question-answering system tailored for the Sri Lankan school curriculum, designed to retrieve curriculum-based answers from structured educational materials. To address the digital divide in rural areas, an offline-accessible version is planned. The study outlines a framework for development and a proposed evaluation using standard NLP metrics and user feedback, aiming to create a scalable, effective tool for Sinhala language learners.
item: Thesis-Full-text
Characterizing meso-mechanical properties of ultra-thin plain woven fibre composites under large deformations
(2025) Weerasinghe, WUD; Mallikarachchi , C; Herath, S
Ultra-thin woven fibre composite structures offer high strength-to-weight ratios and flexibility, making them highly used for aerospace and other applications requiring lightweight, durable materials. However, understanding and predicting the mechani- cal behaviour of these composites under large deformations, particularly their flexural response and bending stiffness, remains a challenge. This study presents a compre- hensive investigation into the meso-mechanical properties of ultra-thin plain woven composites under conditions of extreme curvature. Using finite element modelling, this research focuses on accurately capturing the bending behaviour and assessing the impact of inter-tow slipping and material non-linearity, which are pivotal in under- standing the reduction in bending stiffness observed at high curvatures. A detailed numerical model based on a Representative Unit Cell approach was developed in Abaqus software, incorporating Periodic Boundary Conditions to reflect the composites meso-structural attributes. To simulate realistic conditions, both geo- metrical and material non-linearities were introduced. This approach enabled a robust analysis of cohesive interactions between fibre tows and the resin matrix, which are crucial forreplicating theobservedreductioninbendingstiffnessathighercurvatures. The FE model was calibrated and validated against experimental results from the lit- erature, showing strong agreement with empirical data and confirming the model’s predictive capabilities. Results indicate that interfacial interactions, coupled with material non-linearity, play significant roles in the mechanical response of these composites, particularly un- derlargebendingstrains. Thestudysuccessfullycapturesthesephenomena,providing a framework for further parametric studies and highlighting the necessity of meso- scale modelling for such complex materials. The findings offer valuable insights for optimizing the design of ultra-thin woven composites for use in deployable aerospace structuresandotherapplications,contributingtothedevelopmentofmoreefficientand reliable structural components.
item: Conference-Extended-Abstract
Generative AI for cybersecurity: crafting network intrusion datasets
(Department of Computer Science and Engineering, 2025) Rathakrishnan, M; Gayan, S; Athuraliya, CD
Rapid advances in networking technology and the increasing complexity of cyberattacks have created the need for Artificial Intelligence (AI)-powered Intrusion Detection Systems (IDSs). However, the performance of AI-based IDS is limited due to the lack of labeled intrusion data, class imbalances, and restrictions to share intrusion data due to the General Data Protection Regulation (GDPR). In addition, it is expensive and risky to simulate attacks and collect intrusion data. Recently, Generative Adversarial Networks (GANs) have proven to be promising solutions for synthesizing data. However, most current GAN architectures are designed with Gaussian-like data distributions in mind (such as images) and are difficult to adapt to tabular data characterized by non-Gaussian, mixed-type, and multi-modal features. To address the above-mentioned limitations, we propose a multi-model framework based on Conditional Tabular GAN
(CTGAN) [1] that can be used to synthesize network intrusion data. The framework incorporates rigorous testing and validations to ensure the accuracy and real-world applicability of synthesized intrusion data.
item: Conference-Extended-Abstract
Real-time bus arrival time updating using speed variations
(Department of Computer Science and Engineering, 2025) Farook, S; Thayasivam, U; Athuraliya, CD
Public transportation systems play a critical role in urban mobility, but their efficiency is often hampered by unpredictable delays due to traffic congestion, weather conditions, and other external factors. Accurate real-time bus arrival time predictions are essential to improve the passenger experience, optimize transit operations, and reduce waiting times. Traditional schedule-based estimation methods rely on static timetables, which making them ineffective in dynamic urban environments. This research introduces a data-driven approach of predicting bus arrival in real time using historical and real-time speed variations. By analyzing GPS-based trajectory data and incorporating temporal, spatial, and weather-related characteristics, we aim to improve the accuracy of arrival time estimations.