http://dspace.aiub.edu:8080/jspui/handle/123456789/6 2026-04-02T02:30:20Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2927 Title: Robust Multi-Weather Pothole Detection: An Enhanced YOLOv9 Trained on the MWPD Dataset Authors: Parvin, Shahnaj; Munsy, Foysal; Rahat, Md Tanzeem; Nahar, Aminun; Nur, Kamruddin; Ghose, Debasish Abstract: Real-time pothole detection is crucial for advancing road safety and infrastructure management, particularly in challenging multi-weather conditions. Deep learning-based techniques, especially object detection models, have demonstrated higher accuracy than other approaches. This research proposes an improved YOLOv9 model, specifically designed for detecting road potholes in multi-weather conditions. To optimize performance, ADown layers were replaced with standard convolutional (Conv) layers at specific positions, enhancing feature extraction efficiency while reducing computational load. A custom dataset, the Multi-Weather Pothole Detection (MWPD) dataset, was developed, comprising roadway pothole images captured under varied environmental conditions. Data augmentation techniques, including color perturbation, contrast adjustment, Gaussian noise addition, flipping, and rotation, were applied to enhance training robustness. To ensure a reliable evaluation, a 5-fold cross-validation strategy was employed, partitioning the MWPD dataset into five equal subsets to minimize bias and variance. Using the evaluation benchmarks, the improved YOLOv9 achieved an average mAP@50 of 95% and an F1-score of 91%, outperforming the baseline YOLOv9 model on the MWPD dataset. 2025-10-22T00:00:00Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2926 Title: DeepBERT-XAI: a dual BERT approach with XAI for sentiment analysis of airline tweet data Authors: Rudro, Rifat; Nur, Kamruddin; Sahosh, Zerin; Sneha, Soily; Uddin, Md Hamid; Malik, Sumaiya; Sakib, Fahim; Chowdhury, Rajarshi Roy Abstract: The rapid expansion of social media platforms, particularly Twitter, has transformed how businesses engage in customer sentiments and improve service quality. This study presents DeepBERT-XAI, a hybrid approach that integrates the powerful bidirectional encoder representations from transformers (BERT) architecture with explainable artificial intelligence (XAI) to perform sentiment analysis on 50,000 labeled airline tweets. This study addresses the interpretability of sentiment predictions, providing businesses with actionable insights into customer feedback. Using a dual BERT architecture, the model could effectively process and analyze the language of Twitter posts, accurate sentiment classifications and transparent explanations. The performance of DeepBERT-XAI was assessed using key metrics, and it achieve a training accuracy of 99.00%, validation accuracy of 98.50%, and test accuracy of 98.00%. In addition, it achieved an F1-score of 97.0%, recall of 96.80%, and precision of 97.90%. The significance of this study lies in its context-aware dual BERT fusion and domain-grounded explainability, which uniquely adapts to airline-specific feedback in real time. Unlike static domain-adapted models (AirBERT), DeepBERT-XAI dynamically weights general and domain-specific features via multi-head attention. 2025-12-02T00:00:00Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2925 Title: Recent advancements and challenges of NLP-based sentiment analysis: A state-of-the-art review Authors: Jim, Jamin Rahman; Talukder, Md Apon Riaz; Malakar, Partha; Kabir, Md Mohsin; Nur, Kamruddin; Mridha, MF Abstract: Sentiment analysis is a method within natural language processing that evaluates and identifies the emotional tone or mood conveyed in textual data. Scrutinizing words and phrases categorizes them into positive, negative, or neutral sentiments. The significance of sentiment analysis lies in its capacity to derive valuable insights from extensive textual data, empowering businesses to grasp customer sentiments, make informed choices, and enhance their offerings. For the further advancement of sentiment analysis, gaining a deep understanding of its algorithms, applications, current performance, and challenges is imperative. Therefore, in this extensive survey, we began exploring the vast array of application domains for sentiment analysis, scrutinizing them within the context of existing research. We then delved into prevalent pre-processing techniques, datasets, and evaluation metrics to enhance comprehension. We also explored Machine Learning, Deep Learning, Large Language Models and Pre-trained models in sentiment analysis, providing insights into their advantages and drawbacks. Subsequently, we precisely reviewed the experimental results and limitations of recent state-of-the-art articles. Finally, we discussed the diverse challenges encountered in sentiment analysis and proposed future research directions to mitigate these concerns. This extensive review provides a complete understanding of sentiment analysis, covering its models, application domains, results analysis, challenges, and research directions. 2026-03-01T00:00:00Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2924 Title: XLLC-Net: A lightweight and explainable CNN for accurate lung cancer classification using histopathological images Authors: Jim, Jamin Rahman; Rayed, Md. Eshmam; Mridha, M.F.; Nur, Kamruddin Abstract: Lung cancer imaging plays a crucial role in early diagnosis and treatment, where machine learning and deep learning have significantly advanced the accuracy and efficiency of disease classification. This study introduces the Explainable and Lightweight Lung Cancer Net (XLLC-Net), a streamlined convolutional neural network designed for classifying lung cancer from histopathological images. Using the LC25000 dataset, which includes three lung cancer classes and two colon cancer classes, we focused solely on the three lung cancer classes for this study. XLLC-Net effectively discerns complex disease patterns within these classes. The model consists of four convolutional layers and contains merely 3 million parameters, considerably reducing its computational footprint compared to existing deep learning models. This compact architecture facilitates efficient training, completing each epoch in just 60 seconds. Remarkably, XLLC-Net achieves a classification accuracy of 99.62% 0.16%, with precision, recall, and F1 score of 99.33% 0.30%, 99.67% 0.30%, and 99.70% 0.30%, respectively. Furthermore, the integration of Explainable AI techniques, such as Saliency Map and GRAD-CAM, enhances the interpretability of the model, offering clear visual insights into its decision-making process. Our results underscore the potential of lightweight DL models in medical imaging, providing high accuracy and rapid training while ensuring model transparency and reliability. 2025-05-30T00:00:00Z