Tiziana Cattai

IoT-based ECG monitoring can support well-being assessment across a range of applications, from emotion recognition to arrhythmia detection. In this work, we propose a novel graph–wavelet-based representation of the ECG signal for classification under different emotional and health conditions. The proposed wavelet-graph encodes the persistence of ECG components across multiple wavelet sub-bands, effectively discriminating structurally different signals both in affective computing and arrhythmia detection tasks. The proposed wavelet-graph approach applies a single unified framework across diverse tasks — a characteristic that remains relatively uncommon in the current literature — outperforming task-specific methods by up to 4% and maintaining competitive performance across all evaluated settings, despite those methods being individually optimized for a single classification problem. Furthermore, the proposed approach relies on substantially fewer training samples and a markedly reduced number of trainable parameters compared to existing methods, thereby supporting near–real-time IoT-based wellbeing monitoring applications.

The Dirac operator provides a unified framework for processing signals defined over different order topological domains, such as node and edge signals. Its eigenmodes define a spectral representation that inherently captures cross-domain interactions, in contrast to conventional Hodge-Laplacian eigenmodes that operate within a single topological dimension. In this paper, we compare the two alternatives in terms of the distortion/sparsity trade-off and we show how an overcomplete basis built concatenating the two dictionaries can provide better performance with respect to each approach. Then, we propose a parameterized nonredundant transform whose eigenmodes incorporate a mode-specific mass parameter that captures the interplay between node and edge modes. Interestingly, we show that learning the mass parameters from data makes the proposed transform able to achieve the best distortion-sparsity tradeoff with respect to both complete and overcomplete bases.

Water Distribution Networks (WDNs) are critical infrastructures that require continuous monitoring to ensure efficient operation and mitigate water loss caused by leaks. The strategic placement of sensors within these networks is essential for timely leak detection and localization. Selecting the most effective nodes for sensor placement, however, remains a challenging problem. Centrality measures, widely used in network science to quantify the node importance, offer a promising approach to guide the sensor deployment. This study investigates the impact of various centrality measures on node selection for leak detection in WDNs.We address the leak-detection problem using a two-stage methodology. First, we derive compact and informative feature representations from pressure matrices using the encoder of an autoencoder. Then, we apply a binary classifier to identify anomalous embeddings. We apply our method to two topologically different WDNs, exploring different centrality measures for sparse sensor deployment. Results show that in looped, high-redundancy topologies, a reduced set of strategically placed sensors maintain or even improve detection accuracy. In contrast, more fragile tree-like networks experience performance loss when sensors are reduced. To explain this structural behavior, we introduce a metric that quantifies a network’s structural vulnerability. Taken together, our findings offer practical guidance for water utilities seeking to balance leak-detection speed, accuracy, and cost through topology-based sensor placement.

In EEG-based connectivity analysis, multiple graphs are typically available, as obtained from measurements taken under different conditions, such as frequency bands, trials, or patients' states. Identifying stable patterns of interaction among brain regions, such as functional communities, can provide valuable insight into brain organization and its changes across tasks or conditions. It is then of interest to find a method to properly combine these different graphs to obtain a clustering that is sufficiently stable and representative of brain functionalities. In this paper, we propose a method to obtain robust spectral clustering. The method relies on a statistical characterization of the multiple graphs and a small perturbation analysis of the eigen-decomposition of graphs affected by random perturbations to derive the optimal weighting that minimizes the variance of the eigenvalues. The proposed method is first tested on synthetic data, to assess its advantages with respect to conventional approaches under controllable conditions, and then it is applied to real EEG data in both healthy individuals performing motor imagery tasks and patients affected by Alzheimer's disease. The results show the ability of the method to reliably detect functional communities and quantify connectivity reorganization during cognitive tasks. Our results suggest that the proposed approach provides a valid new strategy to combine multiple graphs taking into account the statistical properties of each graph in the presence of uncertainties.

The interest in applications related to Multi-Unmanned Aerial Vehicle (UAV) systems has been growing exponentially in the last few years. Reinforcement Learning (RL) presents one of the most popular alternatives for solving Multi-UAV tasks, thanks to its flexible requirements for modeling the problem. However, it is often applied to abstractions of the original problem, thus leaving to next development phases the integration of RL solutions to the actual systems. This choice may not guarantee the overall optimal performance of the implemented system. In this survey, we analyze the literature on Multi-UAV applications that utilize reinforcement learning, with particular attention to works that consider realistic communication channels. We focus on identifying the key variables that influence communication and whether these variables are integrated within the RL framework or considered externally. Additionally, we identify key trends, challenges, and future directions in the field, providing a comprehensive overview for researchers and practitioners interested in the practical deployment of RL-based Multi-UAV systems.