BRAC University Institutional Repository
Preserving Knowledge, Advancing Research, Sharing Scholarship
A digital platform for collecting, preserving, and sharing BRAC University’s scholarly, academic, and institutional outputs.
Contribute to the Repository
Faculty members and students are invited to submit their research publications, theses, dissertations, and scholarly works to increase visibility, access, and long-term preservation.
Recent Submissions
Digital literacy and academic performance: The mediating roles of digital informal learning, self-efficacy, and students’ digital competence
(Frontiers Media SA, 2025-01-01) Zakir S.; Hoque, Mohammad Enamul; Susanto P.; Nisaa V.; Alam, Md. Kausar; Khatimah H.; Mulyani E.; BRAC Business School
The rapid evolution of digital technology has significantly influenced learning systems. Digital technology serves as a catalyst for transformative shifts in the manner in which individuals engage in many activities including educational pursuits. The widespread use of digital technology throughout all sectors, including education, has served as a catalyst for students to embrace and utilize new technology. However, challenges arise with the incorporation of digital technology into the classroom. This study investigated the relationship between digital literacy and academic performance, taking into account the role of digital informal learning, self-efficacy, and students’ digital competence as mediators. This study utilized a quantitative methodology employing a structured questionnaire for data collection and Structural Equation Modeling (SEM) for hypothesis testing. This study found that improving students’ digital literacy skills can lead to thriving in academic pursuits. The empirical findings demonstrate that an increase in digital literacy improves digital competence, informal digital learning engagement, and digital self-efficacy. Additionally, possessing digital competence, engaging in digital informal learning, and having digital self-efficacy increases the likelihood of academic success. Therefore, digital competence, digital informal learning, and digital self-efficacy serve as partial mediators in the relationship between digital literacy and academic success. Hence, possessing digital competence, engaging in digital informal learning, and having digital self-efficacy contribute to enhancing the influence of digital literacy on academic achievement. These findings offer insightful implications for educators and policymakers.
Detecting AI-generated paraphrases in Bengali: A comparative study of zero-shot and fine-tuned transformers
(Institute of Electrical and Electronics Engineers Inc., 2025-01-01) Islam M.R.; Samu, Most. Sharmin Sultana; Hossain M.Z.; Zaman F.U.; Bhuiyan M.K.; Department of Computer Science and Engineering
Large language models (LLMs) can produce text that closely resembles human writing. This capability raises concerns about misuse, including disinformation and content manipulation. Detecting AI-generated text is essential to maintain authenticity and prevent malicious applications. Existing research has addressed detection in multiple languages, but the Bengali language remains largely unexplored. Bengali's rich vocabulary and complex structure make distinguishing human-written and AI-generated text particularly challenging. This study investigates five transformer-based models: XLM-RoBERTa-Large, mDeBERTaV3-Base, BanglaBERT-Base, IndicBERT-Base and MultilingualBERT-Base. Zero-shot evaluation shows that all models perform near chance levels (around 50% accuracy) and highlight the need for task-specific fine-tuning. Fine-tuning significantly improves performance, with XLM-RoBERTa, mDeBERTa and MultilingualBERT achieving around 91% on both accuracy and F1-score. IndicBERT demonstrates comparatively weaker performance, indicating limited effectiveness in fine-tuning for this task. This work advances AI-generated text detection in Bengali and establishes a foundation for building robust systems to counter AI-generated content.
Adoption of SPACE-learning management system in education era 4.0: An extended technology acceptance model with self-efficacy
(Frontiers Media SA, 2024-01-01) Sesmiarni Z.; Hoque, Mohammad Enamul; Susanto P.; Islam, Md Asadul; Hendrayati H.; BRAC Business School
The COVID-19 pandemic has caused constraints on education that takes place face-to-face; thus, the learning process has been conducted online. With this emergence, an application-based learning tool called a Learning Management System (LMS) was created to cater to the requirements of distant students. Therefore, this study explores how SPACE-LMS is used in the Teacher Professional Education Program (TPEP) and how it interacts with the Technology Acceptance Model (TAM) and self-efficacy. This study collected data from TPEP participants in the province of West Sumatra-Indonesia from the 2022 cohort, and 245 individuals actively participated. As the quantitative method, Partial Least Square-Structural Equation Modeling (PLS-SEM) analysis with SmartPLS software was used to test and predict the conceptual model. The empirical findings demonstrate that the readiness of adopting SPACE influences perceived ease of use (PEU) and perceived usefulness (PU). Self-efficacy (SE) influences PEU and intention to use (ITU); PEU influences PU; PEU influences attitudes toward users (ATT); and PU influences ATT. Interestingly, PU had no impact on the ITU. The desire of TPEP students to use the SPACE-LMS is also influenced by their sense of self-efficacy. Thus, this study has both theoretical and practical implications.
Rice leaf disease classification: A comparative evaluation of CNNs and vision transformers
(Institute of Electrical and Electronics Engineers Inc., 2025-01-01) Islam, Apu; Rafi, Ishraque Arefin; Rahman, Somaya Al Sadia; Mondal S.; Alam, Md. Golam Rabiul; Department of Computer Science and Engineering
Rice is a staple crop worldwide but is highly vulnerable to leaf diseases that threaten food security. Traditional manual inspection is slow, labor-intensive, and error-prone, highlighting the need for automated detection. This research introduces a comparative analysis between convolutional neural networks (CNNs) and vision transformers (ViTs) for multi-class rice leaf disease classification. Using an 11-class dataset of bacterial, fungal, viral, pest-related, and healthy leaves, we evaluated ResNet50, MobileNetV2, BEiT, DeiT, Swin Transformer V2, and Swin-Tiny V2. Results show transformer models outperform CNNs, with the quantized Swin-Tiny V2 Transformer achieving 99.56% accuracy while reducing model size to 26.89 MB, making it suitable for mobile and IoT deployment. To enhance interpretability and understand the model's decisions, Grad-CAM visualizations highlight important regions, ensuring transparency in model predictions. The findings demonstrate the potential of lightweight, explainable transformers for realworld agricultural IoT applications in precision farming.
Complete implementation of the Green's function based time reverse imaging and sensitivity analysis of reversed time tsunami source inversion
(John Wiley and Sons Inc, 2017-10-16) Hossen, M. J.; Cummins P.R.; Satake K.; Department of Mathematics and Natural Sciences
In recent studies, the Time Reverse Imaging (TRI) method has been implemented in tsunami science to estimate tsunami source models. A TRI algorithm called Green's Function Based Time Reverse Imaging (GFTRI) was previously developed to reconstruct the source model by using source inversion as a guide to appropriately scale time-reversed images of the tsunami source. In this article, we consider a more complete approach to source inversion using reversed time images of the tsunami source by considering cross correlations among Green's functions. As a consequence, the performance of the method improves significantly. We tested this new algorithm using data from the 2011 Japan tsunami. Our results show that the method is capable of extracting more details of the source and providing excellent waveform fits at all stations, including those not used in the source imaging. We have also studied the sensitivity analysis of reversed time tsunami source inversion and found that the method is less sensitive to the number of stations, once a minimum number of stations is utilized. Moreover, this new approach is able to estimate the tsunami source with reasonable accuracy using data available soon after an earthquake, which indicates that it has potential to be used in both near- and far-field tsunami forecasting.