https://ir.kdu.ac.lk/handle/345/7292 Wed, 08 Apr 2026 13:42:01 GMT 2026-04-08T13:42:01Z https://ir.kdu.ac.lk/handle/345/7434 Real-time Taxi Demand and Supply prediction based on Specific Geo locations using Machine Learning - A Systematic Literature Review Abeysekara, AA; Sumanasekara, SG; Samaraweera, SMDD; Imasha, WAC; Mihiran, ASYH; Gunasekara, ADAI; Vidanagama, DU; Lakmali, SMM; Madhubashini, KD; Gunathilaka, HRWP With the increasing popularity of ride-hailing services, the accurate prediction of taxi demand has become a crucial task for service providers. In recent years, the availability of large-scale geospatial data and the development of machine learning algorithms have led to significant advancements in taxi demand prediction.The aim of systematic literature review is to analyze the techniques and approaches for taxi demand and supply prediction using geospatial data and machine learning algorithms.A total of 21 research papers published between 2017 and 2023 were selected based on inclusion and exclusion criteria. The papers were analyzed based on their research objectives,methodology, datasets and evaluation metrics.The result of the literature review indicate that the accuracy of taxi demand prediction models depends on the quality and quantity of the data, the selection of learning algorithms, and the feature engineering techniques used.The systematic literature review highlights the potential of using geospatial data and machine learning algorithms for accurate taxi demand prediction and need for more standardized evaluation metrics and further research to address the challenges. Machine learning algorithms, such as linear regression, decision trees, and artificial neural networks, clustering have been widely used for prediction tasks, focusing on factors like real-time population data.Through a comprehensive analysis, it is determined that clustering emerges as the most suitable technique for the research. Fri, 01 Sep 2023 00:00:00 GMT https://ir.kdu.ac.lk/handle/345/7434 2023-09-01T00:00:00Z https://ir.kdu.ac.lk/handle/345/7433 Advancements in Breast Cancer Detection: Exploring Machine Learning Techniques for Accurate Diagnosis and Early Detection Munasinghe, MASD; Samaraweera, WJ One of the most prevalent illnesses affecting women worldwide is breast cancer. It increases in countries where the majority of cases are discovered in the late stages. The machine learning (ML) technique that is used in this paper to detect breast cancer is retrieved from a digitized mammogram image. It Aimed to evaluate and compare the performance of various machine-learning algorithms such as Convolutional Neural Networks (CNN), Random Forest, Support Vector Machine (SVM), Logistic Regression, and K-Nearest Neighbors (KNN) for breast cancer detection. Using a comprehensive dataset of "RSNA Screening Mammography Breast Cancer Detection", these mammographic images and clinical information are divided into training and testing phases to implement the ML algorithms. The objective was to determine which algorithm yielded the highest accuracy in predicting breast cancer, as this is a critical factor in early detection and successful treatment. research highlights the Convolutional Neural Network (CNN) gives 95.2% accuracy as the most effective machine learning algorithm for breast cancer prediction. CNN’s ability to learn intricate patterns from mammographic images and its superior accuracy make it a valuable tool in early breast cancer detection. These findings have significant implications for improving patient outcomes and the overall effectiveness of breast cancer screening and diagnosis. CNNs revolutionize computer vision, enabling accurate breast cancer diagnosis and detection through automatic learning and feature identification in medical imaging tasks. website's backend will employ the algorithm that produces the best results, and the model will categorize cancer as benign or malignant. Fri, 01 Sep 2023 00:00:00 GMT https://ir.kdu.ac.lk/handle/345/7433 2023-09-01T00:00:00Z https://ir.kdu.ac.lk/handle/345/7432 Application of Artificial Intelligence in Prosthetics: A Review Geethanjana, HKA; Hettige, B This review paper explores the application of artificial intelligence (AI) in advanced prosthetic devices, including limbs, retinal prosthetics, hearing prosthetics, and ortho dental prosthetics, with the aim of enhancing functionality and customization. The research problem centers around understanding AI's forthcoming impact on prosthetic advancements. The study's objectives are twofold: to identify current AI applications in prosthetics and to project future possibilities. The paper uses qualitative secondary analysis to review existing research. By leveraging AI algorithms, prosthetic limbs can interpret nerve signals derived from the patient's muscles, resulting in more precise control and operation. AI-driven advancements include myoelectric prostheses that utilize electromyography signals, bionic legs that adapt to different environments based on user feedback, and prosthetic arms capable of executing actions using computer vision recognition. Additionally, AI improves retinal prosthetics by combining neural networks with computer vision techniques to refine facial features, enhance environmental representation, and ensure safety. In hearing prosthetics, AI, machine learning, and neural networks enable devices to adapt to individual hearing needs and background noise environments. AI-based object detection techniques streamline dental implant surgery in ortho dental prosthetics. The integration of AI in prosthetic devices holds the potential to enhance functionality, improve control and customization, and provide a more natural user experience, benefiting millions worldwide with limb amputations, vision and hearing impairments, and dental prosthetic needs. Fri, 01 Sep 2023 00:00:00 GMT https://ir.kdu.ac.lk/handle/345/7432 2023-09-01T00:00:00Z https://ir.kdu.ac.lk/handle/345/7431 A Web-based Learning system for the Japanese Language a.Proficiency Test in Sri Lanka Rajapaksha, KC; Kathriarachchi, RPS; Wijesinghe, PRD The Japanese Language Proficiency Test (JLPT) is a standardized exam that assesses nonnative speakers' understanding of the language as well as their reading, writing, and listening skills. The majority of people in Sri Lanka take JLPT to obtain the required qualifications to study and work in Japan and improve their knowledge of the language. However, they have fewer requirements to gather the information and proceedings about JLPT. Due to its complexity and numerous letter patterns with varied meanings, the Japanese language is challenging to learn for Sri Lankans. The purpose of this research is to investigate the issues that occur to people when they are following JLPT. As primary data, this study mainly focuses on the survey that was conducted through social media platforms from the participants from those who have already completed JLPT. This used statistical methods which were quantitative and used published research studies related to the research area as secondary data which are qualitative. According to the survey results, the researcher identified the main problems that responders faced when they did the JLPT. To overcome these problems, this paper proposed a web-based Learning system with the Japanese bot as an assistant for those who are taking JLPT, and this describes the main functions that have to be included for this proposed system and how it could be helpful for the people who are doing this examination. As a future avenue, this system could be implemented as a multi-language system with the J-bot for the other language examinations in Sri Lanka. Fri, 01 Sep 2023 00:00:00 GMT https://ir.kdu.ac.lk/handle/345/7431 2023-09-01T00:00:00Z