https://ar.iub.edu.bd/handle/11348/979 2026-04-16T04:53:40Z https://ar.iub.edu.bd/handle/11348/1039 A Comprehensive Hierarchical Review of Blockchain Technology in Revolutionizing Carbon Emission Trading System Khan, Nayeem; Tahmid, Ahanaf; Monir Adar, Ahasanul The growing climate crisis has raised the need to search for reliable mechanisms which could reduce greenhouse gas gas emissions worldwide. Carbon markets can continue to play a central role in this effort but often there is a problem of duplication of credit, fraud, multi-jurisdiction and poor traceability. There is an attempt to avoid such shortcomings by suggesting a multi-layered framework, called BAIC-Gov. The framework enables automated and verifiable monitoring and enforcement of carbon credits through blockchain technology, smart contracts, live real-time data on the Internet of Things (IoT) and anomaly detection via Artificial Intelligence (AI). Smart contracts are coded in such a way that they will be directly connected to the emissions data measured by IoT sensors, thus credit issuance as well as compliance will take place autonomously and with precision. Such application was illustrated with the use of the IBM-China ETS and Toucan Protocol. The framework enhances the climate governance, the ability of the market to scale, and the willingness by countries around the world to achieve carbon neutrality to establish a more sustainable future by strengthening confidence and integrity in carbon trading. 2025-08-01T00:00:00Z https://ar.iub.edu.bd/handle/11348/1037 A Unified Lightweight Framework for Detecting Multimodal Attacks Sayem, Md; Rakibul Hasan, Md.; Islam Anika, Morium Modern Vision-Language Model (VLM) has shown tremendous performance in multi- modal reasoning, captioning, retrieval and generative tasks, but it is also fatally suscep- tible to two types of attacks: adversarial perturbation and jailbreak prompts. In order to tackle these balancing threats, the current thesis presents two lightweight, model-free detection systems that will improve the security and resilience of modern multimodal systems. The former contribution is LMDF, which is a semantic-consistency-based ad- versarial detection model to detect perturbation-based attacks by evaluating cross-modal correspondence between image and text embeddings. Based on the conceptual underpin- nings of contrastive learning, LMDF uses the fact that adversarial perturbations, though imperceptible on the eye, introduce quantifiable distortions in the shared embedding space. LMDF identifies adversarial manipulations of frozen pretrained languages encod- ings like CLIP and BLIP-2 with high accuracy through the evaluation of cosine similarity between language and vision encodings. Major experiments between various datasets and attack algorithms (FGSM, PGD and Adversarial Patch) show high effectiveness with maximum accuracy and AUC scores reaching up to 91.2 and AUC up to 0.950 with mini- mum computational costs; only two forward passes and similarity calculations are needed. The second addition is a multimodal jailbreak detecting framework based on confidence which expands the concepts of Free Jailbreak Detection to vision-language environment. This method compares temperature scaled token probability distributions produced by decoder-based VLMs and derives five important statistical properties, namely minimum token confidence, first-token confidence, mean token confidence, entropy and confidence standard deviation. Jailbreak induces the typical instability of these confidence profiles which allow effective classification with a small threshold-based detector. Empirical val- idation shows great discriminative ability with AUC = 0.979, 90 percent accuracy and F1-score = 0.907 at an optimal temperature setting without adjusting its model, gradi- ent access, or evident computing cost.Combined, these two detection modules deal with different yet more and more common multimodal attack vectors. This paper combines semantic alignment analysis of adversarial perturbations with behavioral analysis based on confidence to offer a consistent, practical, and efficient defense mechanism to protect the modern VLMs. The suggested frameworks promote the objective of developing reli- able, multimodal AIs, which can work safely in the real-world environment, high-stakes, and adversarial settings. 2025-12-13T00:00:00Z https://ar.iub.edu.bd/handle/11348/1036 Entropy Regularized Transformer Based - Anomaly Detection Hasan, Mostafa Fahim; Ahmed, Sadia; Hasan, Md. Jahid 2025-12-01T00:00:00Z https://ar.iub.edu.bd/handle/11348/1035 Data-Driven Study of Behavioral Predictors of Myopia: A Machine Learning Based Early Screening Approach Maruf Yasin, Protik; Syeda Ayesha, Mostofa; Marcel Jupiter, Gomes; Shipra, Banik Myopia is growing very fast, particularly in school-going children, primarily owing to lifestyle and educational transformations of digital technology. The research is an evaluation of machine learning models with regards to early and non-clinical prediction of childhood myopia based on behavioral and environmental factors. The 1,002 participants provided a survey of daily screen time, outdoor activity, sleep time, the distance of visiting reading materials, posture and the history of ocular health of parents. Designed to be trained and evaluated using standard metrics of evaluation, the four models, namely: Logistic Regression, Random Forest, Gradient Boosting and a two-layer Artificial Neural Network (ANN) were trained and evaluated. Artificial Neural Network had the largest AUC (0.829) and ANN showed the highest sensitivity with a recall of 96.1, which is why it will be highly suitable in screening settings where false negativity is the major issue. As identified during feature-importance analyses, outdoor exposure, screen-time and duration of sleep were the most significant factors in relation to a risk factor of myopia. Altogether, the results suggest that behavioral data may be successfully used as the support of lightweight and low-cost myopia risk assessment tools, which may be helpful to schools, their parents and primary-care settings in search of the early detection and preventive interventions. 2025-01-01T00:00:00Z