Ensuring consistent quality in friction stir welding (FSW) remains challenging because defect formation and mechanical performance arise from complex thermo-mechanical interactions. This study presents an interpretable convolutional neural network-bidirectional long short-term memory (CNN–BiLSTM) framework for patch-level surface-defect detection and weld-level tensile-strength classification in ultrasonic vibration-assisted FSW AA2060-T8E30 joints. Surface images from 54 welds were divided into six ordered patches to preserve contextual information along the welding direction. DenseNet121 and VGG16 were evaluated as standalone CNNs and hybrid CNN–BiLSTM models. DenseNet121–BiLSTM achieved the best defect-detection performance, with 98.15% fixed-test accuracy and 95.54% mean leave-one-weld-out accuracy, compared with 92.59% and 90.23% for VGG16–BiLSTM. The DenseNet121–BiLSTM model was then applied to three-class tensile-strength classification relative to base-material ultimate tensile strength, achieving 88.89% fixed-test accuracy and 77.78% ± 7.41% under repeated stratified grouped cross-validation. Gradient-weighted Class Activation Mapping (Grad-CAM) visualisations identified surface regions associated with the predictions, while a graphical user interface integrated image input, classification, confidence reporting, and interpretation. The results demonstrate the feasibility of using standard weld-surface images for both visible defect detection and preliminary tensile-strength categorisation to support post-weld decision-making.
An algebraic structure graph is a graph G(V,E) whose vertex set V has an algebraic structure (group, ring, topological space, vector space, etc.). They offer a link between abstract algebra and discrete mathematics, allowing the study of the algebraic properties of discrete structures by graph-theoretic means and the use of graph invariants in information theory, network design, and cryptography. Two novel classes of algebraic structure graphs, square-difference zero-divisor graph (SDZDG) and primitive-rooted graph, are introduced in this work. The vertex set of the SDZDG consists of all the zero-divisors of Zn; two vertices a and b of the SDZDG are connected by an edge if and only if a2≢b2(modn). The vertex set of the primitive-rooted graph is Zn; two vertices a and b are connected by an edge if and only if ab is a primitive root of n. In the case of n=pq (where p and q are two distinct odd prime numbers), SDZDG has been proved to be regular, and the degree sequence, clique number, independence number, and Zagreb indices have been calculated for n=2pq. In the case of the primitive-rooted graph, connectivity has been characterized, the independence number has been determined, and its components have been described.
Ransomware has evolved into a complex, adaptive, and fast-moving adversary category in which static signatures and monolithic classifiers fail to generalise under concept drift, evasion, and behavioural polymorphism. In this paper we present Agentic SABRE (Semantic-Behavioural Arbitration for Ransomware Evaluation): an uncertainty-aware, neuro-symbolic, multi-agent framework for adaptive ransomware detection. SABRE fuses semantic (representation-based) and behavioural (time-window forensic telemetry) evidence, and employs Monte Carlo Dropout inference to quantify epistemic uncertainty for each agent.We introduce a decision-layer orchestrator that performs risk- and uncertainty-aware triage via two interpretable thresholds: a risk score τ and an uncertainty budget κ. High-confidence, high-risk samples are automatically contained, while uncertain or borderline cases are escalated to human analysts, establishing a flexible computational contract between autonomous response and analyst oversight. To support auditability and trust, SABRE integrates post-hoc explainability mechanisms including gradient saliency, permutation importance, and counterfactual analysis, enabling both local and global interpretation of agent decisions.Extensive evaluation on RDset and RanSMAP demonstrates that Agentic SABRE preserves perfect discrimination on saturated semantic datasets (AUC =1.0) while improving robustness under weak behavioural signals, achieving up to a 4.9% relative reduction in false escalations at equal recall and maintaining calibrated predictive uncertainty. Counterfactual analysis further shows that semantic and behavioural decisions can be flipped with bounded perturbation cost, indicating stable and interpretable decision boundaries. Overall, Agentic SABRE is not merely a higher-accuracy detector but an agentic cyber-defence system that combines uncertainty-aware automation, explainable reasoning, and adaptive triage under evolving ransomware threats.
Aim This rapid review was conducted for the NHS England Safer Staffing Programme (2024) England (2024) [1] to evaluate the evidence base for adult critical care nurse staffing.The review addressed four objectives: 1) identifying current staffing decision-making processes; 2) evaluating the impact of staffing models on patient outcomes and service efficiency; 3) assessing effects on burnout and staff retention; 4) identifying effective staffing responses during the COVID-19 pandemic. Methods A rapid systematic review following PRISMA guidelines was conducted across MEDLINE, CINAHL Plus, PsycINFO, and Scopus (2012–2024). Studies examining quantitative, qualitative, and mixed-methods research on registered nurse staffing in adult intensive care were included. Findings Eighty-six studies met the inclusion criteria. Current staffing guidelines predominantly rely on expert consensus (Level 7 evidence) with day-to-day adaptation by senior nurse managers. Emerging empirical data (Level 4–5) suggest an association between nurse-to-patient ratios of 1:1 or higher and reduced mortality and nosocomial infections, though heterogeneity across studies precludes defining an optimal ratio. Workload intensity may be a more sensitive predictor of outcomes than static ratios. High nursing workloads are associated with increased burnout and turnover. Pandemic responses demonstrated that while “upskilling” non-ICU nurses supported surge capacity, it increased the psychological burden on experienced ICU staff. Conclusions While current guidelines rely heavily on expert opinion, there is some evidence from the literature to suggest that moving toward objective, workload-based frameworks might address limitations in current consensus-based models. However, the existing evidence base is largely observational, necessitating further high-quality research to confirm these associations. Implications for Clinical Practice Healthcare organisations should consider integrating objective workload measurement tools to supplement the professional judgement of the Nurse in Charge. Such an approach may provide a more defensible basis for workforce planning and help mitigate the impact of rising patient acuity on staff retention.
Site investigation data indicate that the natural soil properties are inherent spatially variable in three-dimensional (3D) subsurface, and may exhibit strong nonlinear cross-correlation. However, such data are often non-lattice distributed, posing a significant challenge in dealing with the large correlation matrices that arise in 3D non-Gaussian cross-correlated conditional random field simulations. To address this challenge, this study proposes an effective method to efficiently simulate 3D non-Gaussian cross-correlated random fields conditioned on non-lattice data. The proposed method first transforms non-Gaussian correlated incomplete sounding data to independent incomplete sounding data using the inverse transformation of copula model. The resulting incomplete sounding data are then completed via Gibbs sampler, based on which the independent 3D conditional random fields are generated efficiently by exploiting Kronecker product decomposition. The Copula transformation is subsequently used to recover the non-Gaussian cross-correlated 3D conditional random fields. Four bivariate sites with different copula structures, a trivariate site and a pile foundation example are used to demonstrate the proposed method. Results show that the proposed method accurately simulates the 3D non-Gaussian cross-correlated conditional random fields without the need to handle excessively large correlation matrices. It provides an effective tool for high-fidelity multivariate 3D site reconstruction that respects complex nonlinear cross-correlations while simultaneously leveraging observed data to reduce site uncertainty.
