http://hdl.handle.net/1893/2670 This community contains the ePrints and eTheses produced by the Faculty's staff and students. 2026-10-24T14:48:53Z http://hdl.handle.net/1893/38414 Title: Plasticity in a bacterial global regulatory switch that drives a shift in antibiotic resistance and virulence Author(s): Correia, Annapaula; Manners, Emma J; Evans, Benjamin A; Malone, Jacob G; O’Grady, Justin; Mavrogiorgou, Eleni; Arminu, Sergio; Usai, Manuela; Naik, Dixita; McMillan, Stuart; Kay, Gemma L; Hill, Claire; Sudhakar, Padhmanand; Meader, Emma; Desbois, Andrew P Abstract: Antibiotic resistance and expression of virulence factors impact the outcome of infection by Pseudomonas aeruginosa. Pathogenesis is often modelled using the PAO1 reference strain but laboratory lineages vary in the sequence and activity of MexT, a global regulator impacting virulence, biofilm formation, and ciprofloxacin resistance. We defined the impact of active versus inactive MexT in PAO1 and observed transcriptomic changes affecting the expression of ~900 genes. Phenotyping revealed altered metabolism, antibiotic resistance, and virulence, resulting in striking variation across a ‘single’ model organism. We propose that antibiotic resistance promotes plasticity in mexT accounting for variation across lineages. We introduced antibiotic resistance into clinical P. aeruginosa isolates and observed mutations in mexT when selective pressure was removed, supporting the proposed evolutionary pathway. Overall, we have demonstrated the transcriptomic basis of MexT as a phenotypic switch in PAO1 and implicated antibiotic resistance as a cause of changes in mexT. Furthermore, MexS/MexT-regulated efflux is implicated in the antibiotic stress response and virulence, helping identify the mechanisms for rapid phenotypic switching through mexT and confirming that PAO1 is unlike most isolates. Improved understanding of the regulatory changes linked to antibiotic resistance is particularly relevant to P. aeruginosa where cycles of antibiotic treatment are common. 2026-01-08T00:00:00Z http://hdl.handle.net/1893/38412 Title: Revealing potential Atlantic salmon bycatch hotspots from pelagic fisheries landings across the Northeast Atlantic Author(s): Borland, Aislinn; Elliott, Sophie; Gillson, Jonathan P; Utne, Kjell; Banas, Neil; Toms, Simon; Tyldesley, Emma; Bárðarson, Hlynur; Bull, Colin Abstract: Atlantic salmon (Salmo salar) marine survival has declined markedly over the past five decades, leading to concerns about future population viability. While many studies have explored the impacts of stressors such as climate change and predation, less is known about bycatch from pelagic fisheries targeting finfish species. This study, focused on the first few months of the salmon marine migration period, narrows down the space-time domains where pelagic fisheries bycatch is most likely to occur. Landings from an International Council for the Exploration of the Sea led data call were used to assess bycatch exposure. Temporal patterns of monthly landings of six commercially important pelagic finfish species (mackerel Scomber scombrus, herring Clupea harengus, horse mackerel Trachurus trachurus, sardine Sardina pilchardus, capelin Mallotus villosus, and chub mackerel Scomber colias) within the Northeast Atlantic region from 2000 to 2022 were considered alongside reconstructed salmon migration regions. This study indicated that bycatch exposure from herring and mackerel fisheries was higher than that of other pelagic fisheries along the shared salmon migration corridor between April and August, with cumulative exposure highest over time for the herring fishery. Southern salmon populations were also exposed to the horse mackerel and sardine fisheries in April. Temporal patterns of monthly pelagic landings were compared with patterns of marine return rates in salmon index populations, but no consistent relationships were present. Quantifying salmon bycatch rates at sea remains challenging due to highly variable and sparse data but, given the severity of declines across their range, conservation measures that address factors affecting marine survival of salmon are urgently needed. Future investigations using emerging technology in the pelagic fisheries identified in this study could provide opportunities to develop more robust information on bycatch risk to Atlantic salmon stocks and to direct appropriate mitigation measures. 2026-10-15T00:00:00Z http://hdl.handle.net/1893/38410 Title: An Explainable Estimation Model for Financial Time Series Author(s): Takahashi, Osamu; Cuyt, Annie; O'Hare, Anthony; Lee, Wen-shin; Ferrara, Massimiilinao Editor(s): Vantaggi, Barbara; Coletti, Giulianella; Denoeux, Thierry; Laurent, Anne; Petturiti, Davide; Miranda, Enrique; Medina, Jesus; Bouchon-Meunier, Bernadette; Yager, Ronald R Abstract: Recent developments in artificial intelligence have led to powerful machine learning approaches to model fluctuating dynamics in financial time series, even in the presence of non-stationary behavior. However, such methods often suffer from limited interpretability while uncertainty, arising from complex data fluctuations, remains difficult to quantify. We present a method that decomposes financial data into different frequency components and trends, thereby enabling explicit identification of cyclical patterns and long-term trends. Concurrently, noise components are separated from signal components. Its effectiveness and its high predictive accuracy have been demonstrated in the interpretation of some financial time series. 2026-06-15T00:00:00Z http://hdl.handle.net/1893/38409 Title: Beyond Frame‐Level Accuracy: Analyzing Video‐Level Stability in Transformer‐Based Deepfake Forensics Author(s): Hasan, Syed Azmul; Hera, Mowmita Pervin; Mim, Afroza Lona; Jihad Hossain, M M; Ali, Hazrat; Bulbul, Mohammad Farhad Abstract: Deepfake videos generated by advanced generative models pose serious threats to digital security, misinformation control, and media forensics. Although numerous detection methods have been proposed, their reported performances are often based on heterogeneous preprocessing pipelines, inconsistent data splits, and model-specific training protocols, hindering fair architectural comparison. This work presents a controlled and reproducible comparative study of convolutional neural networks (ResNet-18 and EfficientNet-B0) and a Transformer-based architecture (ViT-B/16) within a unified, face-centric deepfake detection framework. All models are trained using identical preprocessing, augmentation, optimization settings, and data splits on Celeb-DF v2 and FaceForensics++, and results are validated across three independent random seeds with bootstrapped confidence intervals. Experimental results show that while CNN-based models perform competitively at the frame level, their performance degrades after video-level aggregation. In contrast, ViT-B/16 demonstrates more stable video-level performance, achieving 92.8% (95% CI: 91.7–94.0) video-level accuracy and an area under the curve (AUC) of 0.950 (95% CI: 0.934–0.964) on Celeb-DF v2 (3-seed bootstrapped). Cross-dataset evaluation across the same three random seeds reveals a large, asymmetric generalization gap (AUC 0.79 FF++ ⟶ Celeb-DF vs. 0.64 Celeb-DF ⟶ FF++), indicating that in-domain video-level stability does not transfer to unseen manipulation domains. These findings indicate that global self-attention improves robustness and video-level stability for deepfake detection within a given data domain, while highlighting the need for further work on cross-domain generalization. 2026-10-19T00:00:00Z