SACM - United Kingdom
Permanent URI for this collectionhttps://drepo.sdl.edu.sa/handle/20.500.14154/9667
Browse
54 results
Search Results
Item Restricted Engineering and Classifying Indirect Prompt Injections: A Socio-Technical Approach to AI Agent Security(Saudi Digital Library, 2026) Aldurayhim, Abdulaziz; Abudayeh, MohammadAs autonomous AI agents become more prevalent in enterprise settings, they are now susceptible to "Indirect Prompt Injection (IPI)" attacks, which embed adversarial payloads in external data sources to manipulate the agent's contextual execution and force unauthorised actions. This dissertation holds that IPI attacks constitute a socio-technical phenomenon because they exploit semantic vulnerabilities in agent architectures and psychological manipulation patterns in human communication. To address this threat, the study proposes a three-dimensional taxonomy based on payload type, injection vector, and social engineering trigger. The initial corpus comprised 1,119 instances; following rigorous preprocessing and deduplication, a refined dataset of 858 validated samples was established for modelling. A separate 1,032-sample verified dataset pack was used only as an independent reference for dataset verification, not as the definitive modelling dataset. A DeBERTa-v3-base classifier is fine-tuned to classify social engineering triggers, using RoBERTa-base as a baseline. On the OSINT subset, DeBERTa's macro-F1 score was 0.983, and the accuracy was 91.7%. The study also presents a probabilistic Confused Deputy threat model and an ACE defence architecture consisting of architectural controls, runtime detection, privilege minimisation, human-in-the-loop oversight, and practitioner education. The findings conclusively demonstrate that integrating advanced NLP detection mechanisms with socio-technical security frameworks significantly enhances enterprise resilience against IPI threats.12 0Item Restricted Algorithms for Growth Pattern Based Analysis of Lung Adenocarcinoma Histology(Saudi Digital Library, 2026) AlRubaian, Arwa; Rajpoot, Nasir; Raza, ShanLung Adenocarcinoma (LUAD) is a leading cause of cancer deaths and shows considerable variation in patient outcomes due to its complex histological, immune, and molecular diversity. Traditional computational pathology methods use global, pattern-agnostic representations that do not capture the structured organization of tumor tissue. This thesis develops new, pattern-aware computational approaches to address this gap. First, we introduce CellOMaps, a biologically informed image representation that encodes the cellular organization of LUAD tissue. By focusing on spatial structure rather than raw visual appearance, CellOMaps enables robust, scalable characterization of LUAD histological growth patterns. Using this representation, we propose a framework for dense, patch-level growth pattern classification across Whole Slide Images (WSIs), offering a reproducible and spatially resolved alternative to conventional predominant-pattern assessment and demonstrating improved robustness across institutions. Next, we investigate immune heterogeneity in LUAD by analyzing the spatial distribution of Tumor-Infiltrating Lymphocytes (TILs). We introduce GPS-TILs, a growth pattern–specific digital biomarker that quantifies the presence, density, abundance, and spatial dispersion of immune infiltration within distinct morphological patterns. This pattern-aware immune profiling improves prognostic stratification compared to global TILs assessment and manual grading, highlighting the importance of incorporating histological context into immune analysis. Finally, we address molecular inference from histopathology by predicting Tumor Mutational Burden (TMB) from WSIs. We introduce GRIL-GNN, a graph-based learning framework that aggregates information across spatially organized tissue regions while preserving detailed morphological features. This approach uses a novel unified learning objective that combines supervised learning with self-supervised regularization, enabling robust representation learning under weak supervision. By restricting analysis to localized tissue regions with shared histological patterns, this work examines the distribution of the TMB predictive signal within different morphological patterns. Overall, this work demonstrates that integrating spatial and pattern-aware analysis improves the interpretation of morphological, immune, and molecular signals in LUAD. The approaches proposed here provide a framework for more precise, clinically relevant computational pathology, with potential applications to other cancers characterized by complex tissue structure.14 0Item Restricted REGULATING ALGORITHMIC DISCRIMINATION UNDER THE EU AI ACT: EVALUATING BIAS MITIGATION DUTIES FOR HIGH-RISK AND GENERAL-PURPOSE AI SYSTEMS(Saudi Digital Library, 2025) ALSOMALI, ABDULAZIZ; ZIHAO, LIAlgorithmic systems now allocate work, credit, welfare and even police attention (facial recognition systems). They are not ‘neutral’ instruments; they often reproduce and amplify structural disadvantage. This dissertation asks whether the European Union’s Artificial Intelligence Act, when coupled with the Charter of Fundamental Rights and the equality acquis, can prevent and redress such discrimination. This dissertation argues that the Act is normatively necessary but only conditionally sufficient. Its risk architecture, data‑governance duties, documentation and oversight requirements, and the upstream regime for general‑purpose models supply the right legal levers. Constitutional adequacy will materialise only if implementation embeds equality law into technical practice through three cumulative conditions: (i) standards that require context‑specific metric selection justified by proportionality and the availability of less discriminatory alternatives; (ii) supervision with genuine statistical and legal capacity across the system lifecycle; and (iii) remedial pathways that convert logs and technical files into proof under burden‑shifting rules. Thus this paper turns to a functional comparison with the United States and the United Kingdom shows how adverse‑impact doctrine, discovery, and regulator‑led guidance can be harnessed without sacrificing the coherence of the EU model. Followed by Chapter 5 which sets out a concise implementation blueprint and measurable indicators. On that basis, bias mitigation is framed not as ethics, but as a legal duty by which the Act’s success must be judged.10 0Item Restricted Mitigating Distribution Shift and Label Noise in Deep Neural Network Based Network Intrusion Detection Systems(Saudi Digital Library, 2026) Alotaibi, Fahad Mastor T; Maffeis, SergioDeep Neural Network (DNN)-based Network Intrusion Detection Systems (NIDS) have shown strong performance on offline benchmarks under controlled conditions. However, real-world deployment remains challenging. This thesis focuses on two key challenges: (i) distribution shift, in which benign traffic evolves and attackers continuously devise new strategies; and (ii) label noise, arising from imperfect automated labelling of large volumes of network traffic. To address these challenges, the thesis first provides a critical literature review and develops a structured taxonomy of techniques for handling distribution shift in NIDS and label noise in both NIDS and malware datasets. Building on these insights, it introduces three frameworks. Mateen adapts one-class anomaly detection to evolving benign traffic by combining selective labelling with an ensemble-based mechanism that identifies and responds to shifts with minimal manual effort. Rasd extends multi-class NIDS to detect and integrate newly emerging attack classes, substantially reducing labelling costs through strategic selection of a small, informative, and diverse subset. SLB mitigates label noise by partitioning the dataset into clean and noisy sets and iteratively refining both the model and the labels. Each framework is evaluated extensively across multiple datasets and compared with state-of-the-art baselines. Mateen improves the anomaly-detection F1 score by 4.13% under a light-shift scenario (CICIDS2017) and by 72.6% under an extreme-shift scenario (Kitsune). Rasd increases the novel class detection F1 score by 6.83% on CICIDS2017 and by 19.21% on CSE-CIC-IDS2018. SLB reduces the noise rate in CICIDS2017 (with 30% injected random noise) to below 1.2% and outperforms the vanilla baseline by 11.83% in macro F1. This thesis serves as a reference for researchers and practitioners in cyber security and artificial intelligence. Beyond its literature review and taxonomy, it contributes three frameworks that collectively enhance the robustness of DNN-based NIDS, achieving state-of-the-art results on the evaluated benchmarks.14 0Item Restricted CREATIVE AGENCY ALONGSIDE PERFORMATIVE MACHINES(Saudi Digital Library., 2025) Alkaydi, Wael Abdullah; Sikhwal, RaviThe emotional depth of an artwork is intrinsically linked to the artist’s personal context and journey. However, when generative AI creates art, there’s a prevalent feeling that it can lack this essential human synthesis, resulting in works that may be difficult to resonate with (Grassini & Koivisto, 2024). Risking transitioning towards a language that is difficult for us to relate to meaningfully. The challenge is to develop frameworks that preserve human experience in AI-assisted creative processes, effectively bridging the gap between technological innovation and the emotive power of human articulation (Gutiérrez, 2024). Achieving a delicate balance between balancing responsibility with power (Harris , 2025). As technology has always been interlinked with the evolution of art, this is critical in ensuring that as AI contributes to creative fields, the work retains the depth and authenticity which makes it credible.7 0Item Restricted Leveraging AI Technologies for Advanced IoT Vulnerability Detection(Saudi Digital Library, 2025) Bin Hulayyil, Sarah Hamad N; Shancang, LiThe rapid integration of IoT into smart homes has expanded the attack surface, exposing these environments to increasingly sophisticated cyber and physical threats. Existing security approaches are limited by restricted computational capacity, insufficient transparency in decision-making, poor adaptability to emerging zero-day vulnerabilities, and limited support for end-users. This thesis addresses these gaps by designing, developing, and evaluating a series of lightweight, interpretable, and scalable intrusion detection frameworks tailored to resource-constrained IoT ecosystems. The work follows an experimental, data-driven methodology that combines a critical analysis of current detection techniques with the design, implementation, and evaluation of multiple AI-based models. These include CNNs, domain-adapted large language model architectures such as CyBERT, and multimodal networks that integrate cyber and physical data sources. The models are trained and validated on real-world IoT datasets to assess accuracy, computational efficiency, robustness, and suitability for deployment in IoT ecosystems. The thesis first introduces an explainable detection framework for identifying Ripple20 vulnerabilities, employing feature engineering and interpretable machine learning to improve transparency and user trust. It then advances a featureless detection approach based on large language model architectures, demonstrating that domain-specific models operating on raw byte-level inputs can accurately detect unseen attacks without reliance on handcrafted features. To support practical deployment, an accessible detection interface is developed, enabling both expert and non-expert users to analyse network traffic and receive mitigation guidance. Finally, a multimodal intrusion detection framework is proposed that fuses network traffic with video data, enhancing situational awareness and improving detection performance in cyber-physical settings. Collectively, these contributions address the core challenges of explainability, scalability, lightweight operation, usability, and multimodal analysis, thus extending the understanding of how advanced deep learning and language-based models can be applied to IoT security and outline directions for future research on deployable, user-centred intrusion detection in smart home environments.13 0Item Restricted AI for Fraud Detection(Saudi Digital Library, 2026) Albaqami, Abdullah; Muneeb, AhmadFinancial fraud is rapidly growing in the digital payment systems, and it highlights the shortcomings of the fixed rule-based controls and machine learning models that work well in the testing environment but fail miserably in the real-life operational environment. This study develops and evaluates a complete fraud detection pipeline designed to address three persistent challenges: severe class imbalance, model instability under shifting data distributions, and the need for transparent decision outputs required by regulators and financial institutions. The pipeline integrates systematic data preprocessing, an optimized LightGBM model, and SHAP-based interpretability using the IEEE-CIS dataset of 590,540 transactions. The methodology includes memory optimization, structured missing-value treatment, outlier handling through winsorization, label encoding for high-cardinality categorical fields, temporal feature engineering, and correlation-based feature reduction. Optuna is a Bayesian optimisation that is used to optimise LightGBM hyper-parameters using ROC-AUC as the objective function. ROC-AUC, PR-AUC, precision, recall, F1-score, and a confusion matrix are used to measure model performance, thus, following the best practices in imbalanced classification. SHAP analysis is used to produce both global and local explanations of model behaviour. The final model achieves strong discriminative performance, with a ROC-AUC of 0.9606 and a PR-AUC of 0.8042. The accuracy (0.7335) and recall (0.7491) indicate balanced detection and the confusion matrix shows that there is good fraud detection with controllable false-positives. SHAP analysis shows that count based features, transaction amount, card identifiers, geographic features, and temporal patterns are the predictive features, which are consistent with the established fraud behaviours reported in the recent literature. The results demonstrate that the improvement in performance is not only due to the choice of the model but also to the mutual complementary effect of data engineering, hyper-parameter optimization, and interpretability. The researchers conclude that an end-to-end pipeline improves the accuracy of detection, increases transparency, and overcomes fundamental limitations that were found in previous studies of fraud. Limitations are anonymisation of the datasets, lack of drift analysis, and possible loss of fraud indicators in the course of preprocessing.13 0Item Restricted Evaluating Static, Contextual, and End-to-End Embedding Techniques for Malware Detection on Dynamic API Call Data(Saudi Digital Library, 2026) Basfar, Mohammed Raed; Joey, LamThe rate of malware development continues to challenge cybersecurity, with traditional signature- and heuristic-based techniques overwhelmed by polymorphic and zero-day attacks. Natural language processing (NLP) offers a promising direction by modeling dynamic API call sequences as semantic linguistic data, enabling sophisticated embedding and sequence-learning methods to be used for malware detection. This dissertation contrasts and analyzes three typical embedding methods static, contextual, and end-to-end task-learned representations—under a shared experimental framework. Specifically, it employs Word2Vec embeddings with a Convolutional Neural Network (CNN), contextual BERT embeddings with a CNN, and a Bidirectional Long Short-Term Memory (BiLSTM) network with a trainable embedding layer and weighted loss function to address class imbalance. The experiments were conducted on a dynamic API call dataset of around 44,000 malware and 1,000 benign samples, summarized by the first 100 API calls executed under sandboxed conditions. Results indicate that the Word2Vec + CNN pipeline had the highest overall accuracy and malware detection precision but the lowest benign recall. The BERT + CNN model provided more balanced class performance, but at the expense of added computational overhead. The BiLSTM had the highest benign recall, as it was able to easily distinguish from non-malicious activity, but the lowest precision and hugely added resource use. The findings point out the competing trade-offs among detection accuracy, benign recall, and processing efficiency, highlighting the issue of aligning model selection with actual security contexts' resource constraints and priorities. The study contributes by reporting a comparative systematic review of the embedding approaches for malware detection and offering informative insights into performance vs. efficiency trade-offs. Apart from its scientific significance, it proves the larger potential of NLP-based approaches to supporting malware detection systems and to informing the design of responsive, resource-aware cybersecurity systems.20 0Item Restricted Advances in Artificial Intelligence for Energy Forecasting and Performance Management in Buildings(Saudi Digital Library, 2026) Alkhatani, Nasser; Petri, IoanAccurate energy forecasting is essential for intelligent building management, supporting operational optimisation, strategic planning, and demand-side flexibility. However, existing forecasting methods often struggle to remain accurate across multiple time horizons and to generalise across different building types with limited data. This thesis addresses these challenges by developing a modular modelling framework that advances both multi-horizon forecasting and cross-building adaptability. The first contribution is a hybrid forecasting model (SVR → XGBoost → LSTM) designed to deliver stable prediction performance across four horizons: 24 hours, one week, one month, and one year. The hybrid design leverages the complementary strengths of its components SVR for noise reduction, XGBoost for nonlinear feature learning, and LSTM for long-range temporal modelling resulting in improved robustness and generalisation compared with single-model approaches. The second contribution introduces a deep hybrid model (CNN → GRU → LSTM) within a transfer learning framework. Pretrained on multi-building datasets and fine-tuned using limited data from new buildings, this approach enhances cross-domain adaptability while reducing training time and data requirements, demonstrating the practical value of transfer learning for scalable energy forecasting. A third contribution integrates statistical peak detection to support the identification of high- demand events, enabling forecasting outputs to inform grid-interactive building operations. Rigorous evaluation including multi-metric assessment, residual diagnostics, ablation testing, and statistical significance analysis confirms the reliability and robustness of the proposed models. Overall, the thesis provides methodological and empirical advances that strengthen data-driven building energy management. The results show that hybridisation and transfer learning, when carefully designed, can enhance accuracy, stability, and generalisation, offering a scalable pathway toward more efficient and sustainable smart building operations.17 0Item Restricted Artificial Intelligence for Islamic Ontology Linking Qur’an and Hadith Resources(Saudi Digital Library, 2026) Alshammari, Ibtisam Khalaf F; Atwell, Eic; Alsalka, Mohammad AmmarThe primary objective of this thesis is to integrate various Islamic resources, specifically the Qur’an and Hadith, building upon existing academic research. Although previous works have produced valuable contributions, they remain uncoordinated and heterogeneous, raising significant challenges for interoperability and integration. This thesis addresses these limitations by unifying these inconsistent resources into a coherent and comprehensive Islamic ontology, ensuring a more accurate and structured representation of Islamic knowledge and facilitating wider computational applications. The project employs a multi-stage methodology. A workflow combining RML, Cellfie plugin, and SDM-RDFizer interpreter is utilised to integrate Qur’anic resources, enabling data transformation and alignment. This framework successfully maps the data correctly, demonstrating the effectiveness of the Cellfie plugin tool. For the Hadith corpus, knowledge extraction approaches are applied to the LK-Hadith corpus. First experiment with Arabic named entity recognition (ANER) using a limited number of BERT-based Arabic pretrained language models (PLMs) employed on the Sahih Albukhari book produces insufficient results for semantic integration, leading to the adoption of topic modelling methods. The topic modelling and extraction involves Latent Dirichlet Allocation (LDA), Non-negative Matrix Factorization (NMF), BERTopic with Arabic PLMs, and GPT-4. This experiment presented a bilingual Hadith_Teaching_Topics (HTT) dataset. Then, the semantic relationships between Qur’anic and Hadith topics are evaluated through the Quran_Hadith_Dataset (QH_Dataset) using multiple large language models (LLMs), enriching the construction of the Related Quran-Hadith Topics (RQHT) ontology. The ontology is subsequently extended to incorporate the QuranOntology, the entire six Hadith books derived from the LK-Hadith corpus, and their derived topics based on the HTT dataset. This thesis advances the field by establishing a methodological framework for integrating heterogeneous Islamic textual resources by involving multiple data transformations, knowledge extraction, and ontology engineering techniques. Within this framework, the Qur’anic resources are integrated to enhance the unification process, the bilingual HTT dataset is developed to augment the low-resource Arabic annotation and support the final integrated ontology, and the RQHT ontology is introduced to formally represent the semantic relationships between Qur’an and Hadith topics. Collectively, these contributions boost the unification process of Qur’an and Hadith knowledge within a comprehensive and logically consistent semantic Islamic ontology. Evaluation through quantitative and qualitative measures determines the Islamic ontology’s logical consistency, representational completeness, and suitability for advanced computational applications. This research provides a scalable semantic framework to support more knowledge representation and reasoning across core Islamic texts.32 0