Recent Submissions
Evaluation of Machine Learning Models in Student Academic Performance Prediction
(Institute of Electrical and Electronics Engineers Inc., 2025-02-19) Sandeepa A.G.R; Mohottala, S
This research investigates the use of machine learning methods to forecast students' academic performance in a school setting. Students' data with behavioral, academic, and demographic details were used in implementations with standard classical machine learning models including multi-layer perceptron classifier (MLPC). MLPC obtained 86.46% maximum accuracy for test set across all implementations while for train set, it was 99.45%. Under 10-fold cross validation, MLPC obtained 79.58% average accuracy for test set while for train set, it was 99.65%. MLP's better performance over other machine learning models strongly suggest the potential use of neural networks as data-efficient models. Feature selection approach played a crucial role in improving the performance and multiple evaluation approaches were used in order to compare with existing literature. Explainable machine learning methods were utilized to demystify the black box models and to validate the feature selection approach.
Comparative Analysis of Transformer and CNN Architectures for OCT-Based Detection of Circular Leaf Spot Disease in Diospyros kaki
(Institute of Electrical and Electronics Engineers Inc., 2025) Bandara, L.M; Seneviratne, O; Kalupahana, D; Silva, B.N; Wijenayake, U; Wijesinghe, R.E
Circular Leaf Spot (CLS) disease caused by Mycosphaerella nawae poses a significant threat to the cultivation of Diospyros kaki (persimmon). Early and accurate identification is essential to reduce yield loss, yet conventional diagnostic methods are often invasive, subjective, or time-consuming. Optical Coherence Tomography (OCT) offers a non-destructive imaging technique capable of revealing detailed leaf structures, allowing the use of deep learning models for automated disease classification. This study investigates the performance of transformer-based architectures, including ViT, Swin, CoaT, ConViT, LeViT, and MobileViT, for the classification of CLS disease OCT images. The same dataset as in previous CNN-based research was used, applying identical preprocessing techniques for fair comparison. Research was conducted using a 5-fold cross-validation, and transformer models were compared with traditional CNNs from the previous research, such as DenseNet121, VGG16, and Xception. The results show that Transformer architectures, particularly the hybrid CoaT model, achieve superior overall accuracy (87.6%) and micro-averaged AUC (0.974), surpassing the best CNN-based models. The findings highlight the strong potential of transformer architectures in the diagnosis of plant diseases using OCT, especially for early-stage detection. This research underscores the suitability of advanced vision models for high-resolution non-destructive classification tasks in agriculture.
AI Dependency in Sri Lankan Higher Education: Effects of Chatbot Overuse on Student Learning and Problem-Solving
(Institute of Electrical and Electronics Engineers Inc., 2025) Balalojanan, V; Vigneswaran, V; Sukirthan, T; Thayalan, N; Thiruthanigesan, K; Samarakoon, U; Ragel, R. G
Artificial Intelligence (AI) chatbots have rapidly transformed computer science education in Sri Lanka by offering instant coding assistance and conceptual explanations. While these tools enhance learning efficiency, their widespread adoption raises concerns about over-reliance and its impact on students' critical thinking and independent problem-solving abilities. This study investigates AI dependency among 80 computer science students from three Sri Lankan institutions during an open-book programming assignment conducted in May 2025. Students were allowed unrestricted access to AI chatbots and online resources, and their behavior was analyzed using screen monitoring, post-task interviews, and performance assessments. Results revealed that 88% of participants used AI chatbots even for basic tasks, indicating a significant trend of diminished self-reliance. Chatbot-reliant students demonstrated higher accuracy in syntax-based tasks but performed worse in algorithmic design and conceptual articulation. This study introduces the AI Dependency Index (AIDx), a novel metric to quantify the level of reliance across different task types. Findings show that 62% of students fall into the high dependency category, particularly among first- and second-year undergraduates. These insights highlight the need for pedagogical strategies that promote responsible AI usage rather than replace critical thinking and problem-solving skills in higher education.
AI-Driven 3D Avatar Framework for Sign Language Translation and Gesture Representation
(Institute of Electrical and Electronics Engineers Inc., 2025-12-09) Edirisinghe E.A.M.N; Siriwardena A.H.C.D.; Rathnayake R.G.H.M; Sarathchandra H.M.S.D; Jayakody, A; Wijesiri, P
The communication gap between Deaf and Hard-of-Hearing (DHH) communities and mainstream multimedia platforms remains a critical accessibility challenge. Conventional solutions such as subtitles fail to capture the expressiveness, gestures, and finger movements intrinsic to sign language. Existing AI-based avatar systems further suffer from incomplete hand articulation, unnatural body rotations, and desynchronized playback, limiting their realism and effectiveness. To address these challenges, this research introduces AuralFlix, an AI-driven framework that translates spoken or textual content into naturalistic 3D avatar animations for sign language representation. The system integrates MMPose-based pose estimation to capture full-body and hand keypoints, which are mapped to a Mixamo-compatible skeleton in Blender using quaternion-based animation. A custom animation correction pipeline minimizes rotational artifacts, improves finger articulation, and synchronizes the generated motion with video playback through a dedicated multimedia player. Experimental evaluation demonstrates an average positional accuracy of 3.5 mm and rotational error below 3°, achieving approximately 20% improvement in gesture fidelity compared to baseline markerless animation systems. The proposed framework provides a scalable, low-cost, and expressive accessibility solution that enhances digital communication for the DHH community.
Gamified Deep Learning-Based Strategies for Supporting Speech and Motor Skill Development in Children with Autism
(Institute of Electrical and Electronics Engineers Inc., 2025-12-09) Marikkar, U; Mallawaarachchi, T; Senpura, P; Manistar, J; Thelijjagoda, S; Manathunga, K; Vidhanaarachchi, S
Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that affects day-to-day behavior. ASD presents challenges in developing speech, language, and motor skills, which require early intervention. This study proposes deep learning-based gamified techniques to support children with ASD through interactive learning experiences. The system consists of two modules, speech and language assessment and motor skills assessment, addressing both gross and fine motor development. Speech and language capabilities are assessed through analysis of lip patterns and phoneme alignment, using a hybrid model that combines 3D Convolutional Neural Networks (3D-CNN) with Bidirectional Gated Recurrent Units (Bi-GRU), achieving an accuracy of 95%. Fine motor skills are evaluated through the identification of color patterns using a You Only Look Once (YOLO) model combined with geometric methods, achieving a precision of 98%. Gross motor skills are assessed through gesture recognition using Long Short Term Memory (LSTM) networks and angle-based heuristic techniques, achieving 90% accuracy.
The SLIIT Research Document Archive (RDA) is the institutional repository of SLIIT, managed by the SLIIT Library. The primary purpose of SLIIT RDA is to manage, store, and disseminate SLIIT research output with its community and beyond, reaching the wider public. This plays a pivotal role in preserving the academic legacy of the institute.
The collection comprises the research output of SLIIT staff and postgraduate research students, including research publications, conference and symposium papers, books, book chapters, theses, and other scholarly materials. Access to full texts may be restricted depending on the access and licensing terms.