https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4579 2026-04-24T11:24:58Z 2026-04-24T11:24:58Z Ranaweera, P.D.C. https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4717 2023-06-23T04:59:50Z 2023-06-23T00:00:00Z

Title: Lip Synchronization Model for Sinhala Language Using Machine Learning Authors: Ranaweera, P.D.C. Abstract: Currently, a lot of nations produce cartoon characters for a variety of purposes, including the animation film industry, the gaming industry, live broadcast television programs, etc. These characters are made available so that users can interact more with the films, video games, or television shows. For such cartoon figures to appear more alive while speaking a language, lip synchronization is crucial. Lip synchronization is the process of synchronizing speech to a synthetic facial model's lip movement. To create realistic lip-synchronization animation, the voice and lip motions in this procedure must be appropriately timed. Building a talking face utilizing various methods for languages including English, Korean, and Portuguese has been the subject of numerous studies. Compared to other languages, Sinhala has less resources due to less contribution in the researches. The interaction between the synthetic mouth and the Sinhala sounds will be especially interesting to observe. This model can be used to create cartoon characters that speak Sinhala smoothly instead of opening and closing their mouths a lot. The most difficult challenge is to match the "phonemes," which are the fundamental sounds formed in any language, with the "visemes," a visual representation of lip movement. There are three main methods for lip synchronization: the static viseme approach, which uses the viseme alphabet to derive the language's phonemes, the dynamic approach, which employs visual cues from speech in real time, and the deep learning technique, which makes use of a vast visual data set. Because the letters in the Sinhala language indicate the language's phonemes, the viseme classification in this study is based on a variety of letter pairings. Overall, 23 viseme classes have been found. Finally, a deep learning model was produced utilizing a multiclass classification method. In the final system implementation, text input is provided first, after which the system will produce audio and the deep learning model will produce a collection of visemes based on the provided text. The system interface then offers three options for playing the vesmes at various speeds, including rapid, normal, and slow. The user interface was created in Python, and the deep learning model was integrated into the system. The deep learning model for the viseme classification is created using Google collabs. This model will be very helpful in the future when the Sinhala alphabet gets a new character. This approach can also be used to train deaf persons to read lips.

2023-06-23T00:00:00Z Perera, J. A. P. H. https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4716 2023-06-23T04:57:21Z 2023-06-23T00:00:00Z

Title: Multiparameter-based evaluation to identify the trustworthiness of the examinee during e-assessments Authors: Perera, J. A. P. H. Abstract: Monitoring and confirming the examinee's degree of trust during online exams without the assistance of live proctors or invigilators has become a significant concern for most Universities in Sri Lanka. There are currently feasible proctoring options for institutions in Sri Lanka, but they have flaws that make it possible for students to cheat on exams. This study primarily intends to address that issue by introducing a methodology to determine the examinee's trustworthiness during the e-assessment period. As a result, it is critical to comprehend the ideas of trust and trustworthiness in the context of assessment and there are numerous definitions of trust and trustworthiness in the literature. It is essential to highlight that the concept of trust and trustworthiness is context-sensitive, and several metrics are presented in the literature to assess the examinee's identification and behavior-related trustworthiness. Biometric parameters are emphasized in the literature as important factors that are employed for the examinee's identity verification or authentication and there are other non-biometric parameters as well. The term "examinee trustworthiness" as defined in this study refers to the state that an examinee achieves by adhering to the established rules throughout the e-assessment time. Also, researchers have further defined the “trustworthiness parameter” as any action, occurrence, or environmental change that influences the variance in the examinee's level of trust during the online assessment time. In the context of an online e-assessment, this research attempts to construct a multiparameter-based model to measure the examinee's trustworthiness. Using discrete image stream, and operating system event data collected during the online e-assessment, the researcher has conducted experiments and literature analysis in the first stage of this study to identify potential trustworthiness parameters. Through these experiments, the researcher has identified the examinee's environment's unacceptable background, the examinee's lack of visibility in front of the webcam, the presence of multiple people in front of the webcam, the examinee's false identity, and accessing unauthorized materials on the operating system as trustworthiness parameters in this study. Through the parameters used during phase one of this research, the researcher defined the qualitative concept of "examinee trustworthiness" in the context of online e-assessment in a quantitative way. The researcher also determined the proper weights of each chosen parameter to the examinee's trustworthiness because the identified parameters affect the examinee's trustworthiness to varying degrees. The researcher presented the novel concept of the “Trust Index”, which is a quantitative representation of examinee trustworthiness, in an online e-assessment context as an intergraded output of selected parameters as a multiparameter base model to evaluate the examinee trustworthiness. With the help of tooling the researcher had developed to capture each parameter of trustworthiness, the researcher qualitatively assessed the effectiveness of the proposed Trust Model using a simulated e-assessment with 15 participants. The researcher makes many recommendations for tools and the trust model that has been put in place for future research.

2023-06-23T00:00:00Z Gunawardhana, L.G.U.P. https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4627 2022-08-09T15:20:19Z 2022-08-09T00:00:00Z

Title: Optimising Cricket Team Selection for One Day International Series Based on Match Conditions Authors: Gunawardhana, L.G.U.P. Abstract: This thesis focuses on predicting an optimal Sri Lankan cricket team for One Day International (ODI) matches by analysing player performance under different conditions, including weather conditions, opponents and venue. We try to maximise the overall team performance by predicting the best team combination from existing players. The selectors generally perform team selection considering recent performance, including batting and bowling averages of the players. These metrics provide limited insight into players’ potential performance, which leads to drop-ups of qualified. Therefore, consideration of more factors and robust machine learning is required. Our study considers overall performance, consistency, venue, opposition, and recent form of players to predict the players' performance using Random Forest Regression. Then, use the predicted performance to evaluate the player rating of each player towards the team by using Neural Networks. Previous studies have proved that Neural Networks can solve team selection problems successfully [1]. Then we select the team based on the predicted winning contributions to maximise the overall team winning probability. The study concludes by predicting the last 45 matches the Sri Lankan cricket team played during 2017-2019 with the actual playing 11 and the optimal playing 11 selected using our proposed system. We observed that the winning rate of the Sri Lankan cricket team could be improved from 37.77% to 77.77% (105% improvement) if teams were selected using our proposed system.

2022-08-09T00:00:00Z Delgolla, D. M. S https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4626 2022-08-09T15:17:02Z 2022-08-09T00:00:00Z

Title: A predictive model to minimize false positive declines in Electronic Card Not Present financial transactions using feature engineering techniques Authors: Delgolla, D. M. S Abstract: In this research, we have proposed a predictive model to minimize false positive (“Legitimate transactions are being declined falsely identifying as fraudulent”) declines in electronic CNP transactions. Related to the increased popularity of digital payments FP declines are becoming a severe problem among merchants who provide digital payment solutions. It’s estimated that nearly 10% of the transactions are been declined as fraudulent transactions but only very few of them have fallen into the fraud category. To address this problem we have proposed a feature engineering technique based on behavior analysis. Our research is conducted based on a reallife CNP transactional data set from one of the largest fintech service solution providers in Sri Lanka and we have generated 130 features for each transaction and have employed an XG Boost to learn the classifier and found out that performances of the xgBoost classifier has shown nearly 6% improvement in the F-Score and obtained 0.996 for the AUC after the application of feature engineering techniques. We found out that this solution can mainly benefit the merchants who provide electronic payment solutions which involve CNP transactions to minimize false-positive declines targeting legitimate frequent customers and by the same, it minimizes the fraud losses and protects the customer’s interests.

2022-08-09T00:00:00Z