Abstract
Federated learning has emerged as a key paradigm for distributed optimization under privacy and communication constraints, yet classical aggregation schemes such as Federated Averaging (FedAvg) assume uniform client reliability and remain vulnerable to corrupted, noisy, or malfunctioning participants. This limitation is particularly critical in distributed sensing systems, where clients collect partial and heterogeneous observations of a shared physical signal. In this work, we propose a Trust-Aware Federated Signal Reconstruction (TAFSR) framework for robust federated linear inverse problems under client-level corruption. Unlike generic robust aggregators designed for update-space filtering, the proposed method introduces a residual-consistency trust mechanism tailored to inverse problems with partial observations, and admits both a block-wise IRLS interpretation and a spectral-norm error analysis. We formalize federated reconstruction as a distributed least-squares problem with unreliable clients and show that, under strong convexity and weighted-operator spectral bounds, the method converges linearly up to a corruption-dependent bias term. Under persistent residual separation between clean and corrupted clients, the idealized unclipped weighting rule further yields asymptotic attenuation of corrupted-client influence relative to uniform aggregation. The analysis also establishes spectral error bounds and clarifies the connection between trust-aware aggregation and robust M-estimation. Extensive reproducible experiments on industrial vibration data (CWRU) and clinical ECG datasets (MIT-BIH and PTB-XL) under heterogeneous noise and simulated corruption demonstrate that TAFSR consistently improves robustness over FedAvg. Under the default 20% corrupted-client setting, the mean final RMSE decreases from 0.3100 to 0.1999 on CWRU and from 0.2809 to 0.1815 on MIT-BIH, while the method also maintains an advantage on the more challenging low-redundancy PTB-XL regime. These results position TAFSR as a reproducible and task-aware alternative to uniform or generic update-space aggregation for distributed signal reconstruction in heterogeneous sensing environments.

Abstract
Safety-critical interactive systems pose design and evaluation challenges that go beyond usability. The safety of the system (i.e. the guarantee that it does not reach an undesirable or incorrect state) is also a relevant consideration. Traditional user-centred approaches (UCD) lack the rigour and thoroughness needed to address safety, and formal verification arises as a possible solution. Applying formal verification to a safety-critical interactive system design encompasses developing a model, expressing and verifying properties, and analysing the verification results. In the case of model checking, properties are typically expressed in temporal logic. This creates a gap between the languages used in UCD and the languages used for formal verification. Creating temporal logic properties manually requires expertise in formal methods and can be both time-consuming and error-prone. This paper explores how a patterns-based approach can be used to support the specification of properties in a natural language-based style. A prototype implementation of the approach is evaluated through a user study, and the results of this evaluation are discussed.

Abstract
With the rise of Transformers, Vision Transformers (ViTs) have become a new standard in visual recognition. This has led to the development of numerous architectures with diverse designs and applications. This survey identifies 22 key ViT and hybrid CNN-ViT models, along with 5 top Convolutional Neural Network (CNN) models. These were selected based on their new architecture, relevance to benchmarks, and overall impact. The models are organised using a defined taxonomy formed by CNN-based, pure Transformer-based, and hybrid architectures. We analyse their main components, training methods, and computational features, while assessing performance using reported results on standard benchmarks such as ImageNet and CIFAR, along with our training and fine-tuning evaluations on specific imaging datasets. In addition to accuracy, we look at real-world deployment issues by analysing the trade-offs between accuracy and efficiency in embedded, mobile, and clinical settings. The results indicate that modern CNNs are still very competitive in limited-resource environments, while advanced ViT variants perform well after large-scale pretraining, especially in areas with high variability. Hybrid CNN-ViT architectures, on the other hand, tend to offer the best balance between accuracy, data efficiency, and computational cost. This survey establishes a consolidated benchmark and reference framework for understanding the evolution, capabilities, and practical applicability of contemporary vision architectures.

Abstract
This work presents an experimental framework for offshore seismic monitoring that combines Distributed Acoustic Sensing (DAS) with ocean-bottom seismometers (OBS). The study was conducted in the Azores region - Faial, where an HDAS interrogator prototype was connected to dark fiber submarine fiber-optic cable, complemented by the installation of two Ocean Bottom Seismometers (OBS) for calibration and validation of DAS technology. The main objective is to demonstrate that seismic observations obtained by DAS from seafloor cables can provide essential information similar to OBS and particularly in areas where land-based monitoring stations are limited.

Abstract
Waste management networks are essential for environmental protection and public health, but vulnerable to regulatory evasion, fraud, and illegal trading. Detecting potentially illicit activities in the network requires robust anomaly detection systems. However, the complexity of interactions between heterogeneous entities such as recycling companies, individuals, and other organisations combined with temporal irregularities and network dynamics makes conventional fraud detection approaches less effective. In this study, we introduce a dynamic graph-based framework that combines statistical change detection methods, the Page-Hinkley and CUSUM tests, alongside a deep learning model, LSTM-VAE, to detect suspicious activities in the Portuguese waste management network. Using real-world waste transfer records, we engineered temporal and network features to reveal a wide range of anomalies, including abrupt shifts in activity and unusual connectivity patterns, such as those involving collusive triangles. The evaluation was based on four pre-labeled anomalous companies identified by regulators. Our results show that while individual methods excel at detecting certain behaviors, their combination provides robust coverage of diverse anomaly types, with each anomalous company identified by at least three techniques. This approach demonstrates the importance of integrating temporal and network-based analysis, offering regulatory authorities a scalable tool to prioritize inspections, enhancing accountability in the waste management network.

Abstract
The segmentation of the foveal avascular zone (FAZ) in optical coherence tomography angiography (OCTA) images plays a crucial role in diagnosing and monitoring ocular diseases such as diabetic retinopathy (DR) and age-related macular degeneration (AMD). However, accurate FAZ segmentation remains challenging due to image quality and variability. This paper provides a comprehensive review of FAZ segmentation techniques, including traditional image processing methods and recent deep learning-based approaches. We propose two novel deep learning methodologies: a multitask learning framework that integrates vessel and FAZ segmentation, and a conditionally trained network that employs vessel-aware loss functions. The performance of the proposed methods was evaluated on the OCTA-500 dataset using the Dice coefficient, Jaccard index, 95% Hausdorff distance, and average symmetric surface distance. Experimental results demonstrate that the multitask segmentation framework outperforms existing state-of-the-art methods, achieving superior FAZ boundary delineation and segmentation accuracy. The conditionally trained network also improves upon standard U-Net-based approaches but exhibits limitations in refining the FAZ contours.