Research Publications
Permanent URI for this communityhttps://rda.sliit.lk/handle/123456789/4194
This main community comprises five sub-communities, each representing the academic contribution made by SLIIT-affiliated personnel.
Browse
62 results
Search Results
Item Embargo HCLIP: Beyond CLIP for Cost-Effective Multimodal Retrieval in Education(Institute of Electrical and Electronics Engineers Inc., 2025-12-09) Weerasinghe, S; Gunatunga, O; Dewpura, W; Fernando, S; Kasthurirathna, D; Rathnayake, SMultimodal retrieval systems have gained significant attention due to their ability to process and cross-retrieve data containing images and text. However, the factors such as high cost of development, limitation on resources, and the proper addressing of the modality gap, the inherent representational differences between modalities pose a challenge to building effective and efficient retrieval models. In this work, we propose a low-resource, cost-efficient hybrid multimodal retrieval model that integrates Contrastive Language-Image Pre-training (CLIP) and All-MiniLM-L6-v2 to create a shared embedding space while storing raw images in an unstructured database. Our primary contributions include (1) the development of a hybrid model that outperforms CLIP-native retrieval, (2) a novel bidirectional neural network alignment technique that brings textual and visual modalities closer together, and (3) a comprehensive analysis of the modality gap's impact on downstream retrieval performance. Through proper evaluation using transparent techniques such as Mean Reciprocal Rank (MRR) and Cosine-Weighted MRR, our method demonstrates improved retrieval accuracy over baseline approaches. Experimental results exhibit that a lower modality gap does not always prove to be efficient on the downstream retrieval. Our findings pave the way for more efficient, adaptable, and cost-effective multimodal retrieval methodologies in low-resource environments, not limited to the education domain.Item Embargo Explainable Multilingual Sentiment Analysis for Sinhala, English and Code-Mixed Banking Reviews(Institute of Electrical and Electronics Engineers Inc., 2025-12-09) Senevirathna, P; Adhikari, N; Navojith, T; Rizvi, A; Kasthurirathna, D; Abeywardhana, LThe growth of multilingual and code-mixed communication in the financial sector presents significant challenges for sentiment analysis, particularly in low-resource languages such as Sinhala. Existing NLP models often underperform due to limited annotated data, domain-specific vocabulary gaps, and a lack of transparency. This paper presents a multilingual and explainable sentiment analysis framework that addresses the challenges of low-resource and code-mixed languages by analyzing banking customer feedback in English, Sinhala, and Sinhala-English code-mixed formats. A dataset of 13,000 aspect-tagged banking reviews spanning five key banking aspects is used for evaluation. The framework classifies comments into positive, neutral, or negative sentiment and provides interpretable explanations using SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model Agnostic Explanations). English sentiment classification is performed using a fine-tuned BERT-base-uncased model, while Sinhala/code-mixed sentiment is handled by a hybrid XLM RoBERTa with a domain-specific lexicon correction approach. The framework achieves 92.3% accuracy and 0.89 F1-score for English, and 88.4% accuracy and 0.84 F1-score for Sinhala/code-mixed data. The proposed system addresses reproducibility, interpretability, and domain adaptation gaps, offering a deployable solution for multilingual financial sentiment monitoring.Item Open Access Intelligent Systems for Comprehensive Dog Management(Association for Computing Machinery, 2025-06-28) Katipearachchi, M.E; Sachethana, O; Gunawardena, G.N.A; Ruwanara, D.C; Krishara, J; Kasthurirathna, DIn recent years, the integration of advanced technologies with canine welfare has gained significant attention, leading to the development of comprehensive platforms for dog management. The "Research Pooch-Paw"initiative addresses the multifaceted needs of dog owners and stray dog populations through an innovative platform that incorporates machine learning, wearable sensors, and real-time data processing. The platform facilitates early disease detection, behaviour analysis, and health monitoring using IoT-enabled devices, and provides personalized care guidance. Additionally, it includes features for stray dog identification and emergency response using deep learning algorithms and image processing techniques. The research underscores the potential of leveraging modern technology to enhance the quality of life for dogs and improve the effectiveness of canine welfare strategies.Item Embargo Sinhala-English Multilingual AI Call Center Bot with Sentiment-Aware Dialogue and Multimodal CSAT Prediction(Institute of Electrical and Electronics Engineers Inc., 2026-07-06) Kulathunga, T; Amarasinghe, B; Fernando, V; Lakruwani, P; Weerasinghe, M; Kasthurirathna, DThis paper presents a Sri Lanka focused, voice-based call center automation framework supporting Sinhala, English, and Sinhala-English code-mixed conversations in low-resource environments. The system adopts an end-to-end architecture integrating bilingual dataset construction, privacy-preserving speech processing, retrieval-grounded response generation, multimodal sentiment intelligence, and customer satisfaction (CSAT) estimation. A Sinhala-English call-center corpus is created using noise reduction, speaker diarization, and transcript alignment, combined with transcript-aligned PII redaction. During live interaction, language identification routes calls to unified processing pipelines. Real-time sentiment analysis with explainable risk scoring supports escalation decisions, while retrieval-augmented generation ensures factually grounded responses. Emotion-adaptive text-to-speech enhances conversational naturalness. The framework enables interaction-based CSAT estimation without relying solely on post-call surveys, providing scalable and privacy-aware automation tailored to multilingual Sri Lankan call center operationsItem Embargo The Multi-Tenant Customization Paradox: Formalising the Intrinsic Conflict Between Scalable Shared Codebases and Tenant-Specific Operational Customization in SaaS Architecture(Institute of Electrical and Electronics Engineers Inc., 2026-07-07) Jayasuriya, R; Piyarisi, T; Awandya, S; Wickramasooriya, S; Thelijjagoda, S; Kasthurirathna, DMulti-tenant Software-as-a-Service (SaaS) architectures promise economies of scale through a single shared codebase serving multiple tenants. However, enterprise tenants increasingly demand deep operational customization (bespoke workflows, domain-specific business rules, and industry-specific computational logic) that fundamentally conflicts with the sharedcodebase constraint. Despite the centrality of this tension to SaaS architecture, no formal definition exists in the literature. This paper formally defines the Multi-Tenant Customization Paradox (MTCP): the structural impossibility of simultaneously maximizing codebase unity and tenant customization depth without incurring costs that grow super-linearly with the number of tenants. We introduce a formal model quantifying this tension through the Paradox Coefficient $P(S)$, establish a five-dimensional customization taxonomy with per-dimension formal measures, derive the Feasibility Region governed by an architectural sophistication parameter $α(B)$, propose an operational rubric for estimating $α(B)$ in practice, and derive an upper bound on the achievable unity-depth trade-off for four canonical resolution strategies. Our formalization demonstrates that the paradox is inherent to the mathematical structure of multi-tenancy rather than incidental to implementation choices, providing architects and researchers with a theoretical foundation for reasoning about customization trade-offs.Item Embargo Hybrid Model-Based Automated Exterior Vehicle Damage Assessment and Severity Estimation for Insurance Operations(Institute of Electrical and Electronics Engineers Inc., 2025) Jayagoda, N.M; Kasthurirathna, DAfter a vehicle accident, insurance companies face the critical task of assessing the damage sustained by the involved vehicles, a process essential for maintaining the insurer's credibility, building consumer trust, and meeting legal and ethical obligations. This assessment is crucial for ensuring clients' financial protection and proper compensation, upholding the integrity of the insurance process. Traditionally, evaluations have been conducted through manual inspections by experienced professionals who meticulously document vehicle damage. Despite its thoroughness, this approach suffers from significant inefficiencies, high costs, and extended time requirements. Moreover, the method is vulnerable to human errors and subjective biases, which can result in inflated valuations. To overcome these challenges, this research introduces an innovative system designed to leverage technology for analyzing images of damaged vehicles uploaded by the user. This system aims to accurately identify the damaged external components, assess the severity of the damage, and determine the repair needs based on the compromised sections of the vehicle. The findings reveal that the hybrid model used in this research is capable of determining vehicle damage severity with an overall accuracy of 73.3%. This level of accuracy demonstrates the model's robust capability to effectively navigate and analyze complex damage patterns, underscoring its practical applications. By accurately determining damage levels on the first assessment, the model reduces the need for further assessments and disagreements, which frequently cause claim delays. This enhancement increases productivity, reduces administrative costs, and improves the customer experience, resulting in a more efficient, transparent, and satisfactory resolution of vehicle insurance claims.Item Open Access Intelligent Systems for Comprehensive Dog Management(Association for Computing Machinery, 2025-06-28) Katipearachchi, M.E; Sachethana, O; Gunawardena, G. N.A; Ruwanara, D.C; Krishara, J; Kasthurirathna, DIn recent years, the integration of advanced technologies with canine welfare has gained significant attention, leading to the development of comprehensive platforms for dog management. The "Research Pooch-Paw"initiative addresses the multifaceted needs of dog owners and stray dog populations through an innovative platform that incorporates machine learning, wearable sensors, and real-time data processing. The platform facilitates early disease detection, behaviour analysis, and health monitoring using IoT-enabled devices, and provides personalized care guidance. Additionally, it includes features for stray dog identification and emergency response using deep learning algorithms and image processing techniques. The research underscores the potential of leveraging modern technology to enhance the quality of life for dogs and improve the effectiveness of canine welfare strategies.Item Embargo Wet-Neuromorphic Computing: A New Paradigm for Biological Artificial Intelligence(Institute of Electrical and Electronics Engineers, 2025-03-31) Perera, J; Balasubramaniam, S; Somathilaka, S; Wen, Q; Li, X; Kasthurirathna, D; Roohi, A; Nelson, M. TAs we delve into a life governed by artificial intelligence (AI), ongoing research continues to discover new forms of intelligence that are efficient and closely mimic an organism’s brain in terms of performance. This article presents a new concept termed wet-neuromorphic computing, in which biological cells or organisms are leveraged to perform computational tasks using their natural molecular functions. We map key neuromorphic properties to natural biological computing observed in bacteria, 3-D organoids, and Caenorhabditis elegans. To expand beyond the inspiration of the brain to create conventional neuromorphic computing, the study presents a case study that demonstrates bacterial AI computing using the gene regulatory neural network derived from Escherichia coli’s gene regulatory network for pattern recognition, validated through wet lab experiments. Finally, challenges and future directions are discussed.Publication Embargo Spatio-temporal graph neural network based child action recognition using data-efficient methods: A systematic analysis(Elsevier Inc, 2025-06-03) Mohottala, S; Gawesha, A; Kasthurirathna, D; Samarasinghe, P; Abhayaratne, CThis paper presents implementations on child activity recognition (CAR) using spatial–temporal graph neural network (ST-GNN)-based deep learning models with the skeleton modality. Prior implementations in this domain have predominantly utilized CNN, LSTM, and other methods, despite the superior performance potential of graph neural networks. To the best of our knowledge, this study is the first to use an ST-GNN model for child activity recognition employing both in-the-lab, in-the-wild, and in-the-deployment skeleton data. To overcome the challenges posed by small publicly available child action datasets, transfer learning methods such as feature extraction and fine-tuning were applied to enhance model performance. As a principal contribution, we developed an ST-GNN-based skeleton modality model that, despite using a relatively small child action dataset, achieved superior performance (94.81%) compared to implementations trained on a significantly larger (x10) adult action dataset (90.6%) for a similar subset of actions. With ST-GCN-based feature extraction and fine-tuning methods, accuracy improved by 10%–40% compared to vanilla implementations, achieving a maximum accuracy of 94.81%. Additionally, implementations with other ST-GNN models demonstrated further accuracy improvements of 15%–45% over the ST-GCN baseline. The results on activity datasets empirically demonstrate that class diversity, dataset size, and careful selection of pre-training datasets significantly enhance accuracy. In-the-wild and in-the-deployment implementations confirm the real-world applicability of above approaches, with the ST-GNN model achieving 11 FPS on streaming data. Finally, preliminary evidence on the impact of graph expressivity and graph rewiring on accuracy of small dataset-based models is provided, outlining potential directions for future research. The codes are available at https://github.com/sankamohotttala/ST_GNN_HAR_DEML.Item Open Access Exploring emergent topological properties in socio-economic networks through learning heterogeneity(2025-12-10) Karavita, C; Lyu, Z; Kasthurirathna, D; Piraveenan, MUnderstanding how individual learning behavior and structural dynamics interact is essential to modeling emergent phenomena in socio-economic networks. While bounded rationality and network adaptation have been widely studied, the role of heterogeneous learning rates–both at the agent and network levels–remains underexplored. This paper introduces a dual-learning framework that integrates individualized learning rates for agents and a rewiring rate for the network, reflecting real-world cognitive diversity and structural adaptability. Using a simulation model based on the Prisoner’s Dilemma and Quantal Response Equilibrium, we analyze how variations in these learning rates affect the emergence of large-scale network structures. Results show that lower and more homogeneously distributed learning rates promote scale-free networks, while higher or more heterogeneously distributed learning rates lead to the emergence of core-periphery topologies. Key topological metrics–including scale-free exponents, Estrada heterogeneity, and assortativity–reveal that both the speed and variability of learning critically shape system rationality and network architecture. This work provides a unified framework for examining how individual learnability and structural adaptability drive the formation of socio-economic networks with diverse topologies, offering new insights into adaptive behavior, systemic organization, and resilience.