arXiv subjects

Pietro Ferraro

Publications and source records attributed to Pietro Ferraro.

At least 19 recordsLinked to original sources

arXiv2026

Explainable deep-learning detection of microplastic fibers via polarization-resolved holographic microscopy

Reliable identification of microplastic fibers is crucial for environmental monitoring but remains analytically challenging. We report an explainable deep-learning framework for classifying microplastic and natural microfibers using polarization-resolved digital holographic microscopy. From multiplexed holograms, the complex Jones matrix of each fiber was reconstructed to extract polarization eigen-parameters describing optical anisotropy. Statistical descriptors of nine polarization characteristics formed a 72-dimensional feature vector for a total of 296 fibers spanning six material classes, including polyamide 6, polyethylene terephthalate, polyamide 6.6, polypropylene, cotton and wool. The designed fully connected deep neural network achieved an accuracy of 96.7 % on the validation data, surpassing that of common machine-learning classifiers. Explainable artificial intelligence analysis with Shapley additive explanations identified eigenvalue-ratio quantities as dominant predictors, revealing the physical basis for classification. An additional reduced-feature model with the preserved architecture exploiting only these most significant eigenvalue-based characteristics retained high accuracy (93.3 %), thereby confirming their dominant role while still outperforming common machine-learning classifiers. These results establish polarization-based features as distinctive optical fingerprints and demonstrate the first explainable deep-learning approach for automated microplastic fiber identification.

physics.optics↗

arXiv2025

Machine Intelligence on the Edge: Interpretable Cardiac Pattern Localisation Using Reinforcement Learning

Matched filters are widely used to localise signal patterns due to their high efficiency and interpretability. However, their effectiveness deteriorates for low signal-to-noise ratio (SNR) signals, such as those recorded on edge devices, where prominent noise patterns can closely resemble the target within the limited length of the filter. One example is the ear-electrocardiogram (ear-ECG), where the cardiac signal is attenuated and heavily corrupted by artefacts. To address this, we propose the Sequential Matched Filter (SMF), a paradigm that replaces the conventional single matched filter with a sequence of filters designed by a Reinforcement Learning agent. By formulating filter design as a sequential decision-making process, SMF adaptively design signal-specific filter sequences that remain fully interpretable by revealing key patterns driving the decision-making. The proposed SMF framework has strong potential for reliable and interpretable clinical decision support, as demonstrated by its state-of-the-art R-peak detection and physiological state classification performance on two challenging real-world ECG datasets. The proposed formulation can also be extended to a broad range of applications that require accurate pattern localisation from noise-corrupted signals.

eess.SP↗

arXiv2025

Learning Network Dismantling Without Handcrafted Inputs

The application of message-passing Graph Neural Networks has been a breakthrough for important network science problems. However, the competitive performance often relies on using handcrafted structural features as inputs, which increases computational cost and introduces bias into the otherwise purely data-driven network representations. Here, we eliminate the need for handcrafted features by introducing an attention mechanism and utilizing message-iteration profiles, in addition to an effective algorithmic approach to generate a structurally diverse training set of small synthetic networks. Thereby, we build an expressive message-passing framework and use it to efficiently solve the NP-hard problem of Network Dismantling, virtually equivalent to vital node identification, with significant real-world applications. Trained solely on diversified synthetic networks, our proposed model -- MIND: Message Iteration Network Dismantler -- generalizes to large, unseen real networks with millions of nodes, outperforming state-of-the-art network dismantling methods. Increased efficiency and generalizability of the proposed model can be leveraged beyond dismantling in a range of complex network problems.

cs.LG↗

arXiv2025

3D Holographic Flow Cytometry Measurements of Microalgae: Strategies for Angle Recovery in Complex Rotation Patterns

Marine ecosystems are in the spotlight, because environmental changes are threatening biodiversity and ecological functions. In this context, microalgae play key ecological roles both in planktonic and benthic ecosystems. Consequently, they are considered indispensable targets for global monitoring programs. However, due to a high spatial and temporal variability and to difficulties of species identification (still relying on microscopy observations), the assessment of roles played by these components of marine ecosystems is demanding. In addition, technologies for a 3D assessment of their complex morphology are scarcely available. Here, we present a comprehensive workflow for retrieving 3D information on microalgae with diverse geometries through holographic microscopy operating in flow-cytometry mode. Depending on the rotation patterns of samples, a tailored approach is used to retrieve their rolling angles. We demonstrate the feasibility of measuring 3D data of various microalgae, contingent to the intrinsic optical properties of cells. Specifically, we show that for quasi-transparent and low-scattering microorganisms, the retrieved angles permit to achieve quantitative 3D tomographic Refractive Index (RI) mapping, providing a full characterization of the alga in terms of its inner structure and the outer shape. Moreover, even in the most challenging scenarios, where microalgae exhibit high light absorption or strong scattering, quantitative 3D shape reconstructions of diatoms and dinoflagellates can be at least achieved. Finally, we compare our direct 3D measurements with 2D inferences of 3D properties, obtained using a commercially available microscopy system. The ability to non-invasively obtain 3D information on microalgae marks a fundamental advancement in the field, unlocking a wealth of novel biological insights for characterizing aquatic ecosystems.

physics.bio-ph↗

arXiv2025

Inducing, Detecting and Characterising Neural Modules: A Pipeline for Functional Interpretability in Reinforcement Learning

Interpretability is crucial for ensuring RL systems align with human values. However, it remains challenging to achieve in complex decision making domains. Existing methods frequently attempt interpretability at the level of fundamental model units, such as neurons or decision nodes: an approach which scales poorly to large models. Here, we instead propose an approach to interpretability at the level of functional modularity. We show how encouraging sparsity and locality in network weights leads to the emergence of functional modules in RL policy networks. To detect these modules, we develop an extended Louvain algorithm which uses a novel `correlation alignment' metric to overcome the limitations of standard network analysis techniques when applied to neural network architectures. Applying these methods to 2D and 3D MiniGrid environments reveals the consistent emergence of distinct navigational modules for different axes, and we further demonstrate how these functions can be validated through direct interventions on network weights prior to inference.

cs.LG↗

arXiv2024

On Optimal Battery Sizing for Electric Vehicles

In this paper, we introduce a quantitative framework to optimize electric vehicle (EV) battery capacities, considering two criteria: upfront vehicle cost and charging inconvenience cost. For this purpose, we (1) develop a comprehensive model for charging inconvenience costs, incorporating both charging time and detours, improving on existing studies, (2) show, through extensive simulations and analytical models, how charging inconvenience cost is affected by different battery capacity and charging infrastructure configurations, (3) introduce an optimisation framework to determine optimal battery capacities based on charging inconvenience and vehicle cost, and (4) show that optimal battery capacities can be influenced by strategic investments in charging infrastructure and tax/incentive policies. The proposed framework provides actionable insights into the sustainable design of EV systems, supporting the development of cost-effective and convenient electric mobility solutions.

eess.SY↗

arXiv2024

Tree Proof-of-Position Algorithms

We present a novel class of proof-of-position algorithms: Tree-Proof-of-Position (T-PoP). This algorithm is decentralised, collaborative and can be computed in a privacy preserving manner, such that agents do not need to reveal their position publicly. We make no assumptions of honest behaviour in the system, and consider varying ways in which agents may misbehave. Our algorithm is therefore resilient to highly adversarial scenarios. This makes it suitable for a wide class of applications, namely those in which trust in a centralised infrastructure may not be assumed, or high security risk scenarios. Our algorithm has a worst case quadratic runtime, making it suitable for hardware constrained IoT applications. We also provide a mathematical model that summarises T-PoP's performance for varying operating conditions. We then simulate T-PoP's behaviour with a large number of agent-based simulations, which are in complete agreement with our mathematical model, thus demonstrating its validity. T-PoP can achieve high levels of reliability and security by tuning its operating conditions, both in high and low density environments. Finally, we also present a mathematical model to probabilistically detect platooning attacks.

cs.DS↗

arXiv2024

Reinforcement Learning with Adaptive Regularization for Safe Control of Critical Systems

Reinforcement Learning (RL) is a powerful method for controlling dynamic systems, but its learning mechanism can lead to unpredictable actions that undermine the safety of critical systems. Here, we propose RL with Adaptive Regularization (RL-AR), an algorithm that enables safe RL exploration by combining the RL policy with a policy regularizer that hard-codes the safety constraints. RL-AR performs policy combination via a "focus module," which determines the appropriate combination depending on the state--relying more on the safe policy regularizer for less-exploited states while allowing unbiased convergence for well-exploited states. In a series of critical control applications, we demonstrate that RL-AR not only ensures safety during training but also achieves a return competitive with the standards of model-free RL that disregards safety.

cs.LG↗

arXiv2023

Robust decentralised proof-of-position algorithms for smart city applications

We present a decentralised class of algorithms called Tree-Proof-of-Position (T-PoP). T-PoP algorithms rely on the web of interconnected devices in a smart city to establish how likely it is that an agent is in the position they claim to be. T-PoP operates under adversarial assumptions, by which some agents are incentivised to be dishonest. We present a theoretical formulation for T-PoP and its security properties, and we validate this model through a large number of Monte-Carlo simulations. We specifically focus on two instances of T-PoP and analyse their security and reliability properties under a range of adversarial conditions. Use-cases and applications are discussed towards the end of this paper.

cs.DC↗

arXiv2022

On Unique Ergodicity Of Coupled AIMD Flows

The AIMD algorithm, which underpins the Transmission Control Protocol (TCP) for transporting data packets in communication networks, is perhaps the most successful control algorithm ever deployed. Recently, its use has been extended beyond communication networks, and successful applications of the AIMD algorithm have been reported in transportation, energy, and mathematical biology. A very recent development in the use of AIMD is its application in solving large-scale optimization and distributed control problems without the need for inter-agent communication. In this context, an interesting problem arises when multiple AIMD networks that are coupled in some sense (usually through a nonlinearity). The purpose of this note is to prove that such systems in certain settings inherit the ergodic properties of individual AIMD networks. This result has important consequences for the convergence of the aforementioned optimization algorithms. The arguments in the paper also correct conceptual and technical errors in [1].

math.DS↗

arXiv2022

Respiratory Aware Routing for Active Commuters

Cyclists travelling in urban areas are particularly at risk of harm from particulate emissions due to their increased breathing rate and proximity to vehicles. In this paper we combine human respiratory models with models of particulate inhalation to estimate the pollution risk an individual is experiencing in real time given the local pollution level and their heart rate for the first time. Using this model as a baseline, we learn a policy that simultaneously optimises the route for a large number of cyclists with diverse origins and destinations, to minimise overall pollution risk and account for the detrimental impacts of congestion. We learn this policy using reinforcement learning techniques on simulated data in different environments with varying distributions of cyclist fitness. These findings establish that individualised routing is effective in reducing pollution risk while cycling, improving the net benefits of active commuting.

physics.soc-ph↗

arXiv2022

Single-cell phase-contrast tomograms data encoded by 3D Zernike descriptors

Phase-contrast tomographic flow cytometry combines quantitative 3D analysis of unstained single cells and high-throughput. A crucial issue of this method is the storage and management of the huge amount of 3D tomographic data. Here we show an effective quasi lossless compression of tomograms data through 3D Zernike descriptors, unlocking data management tasks and computational pipelines that were unattainable until now.

physics.optics↗

arXiv2022

An adversarially robust data-market for spatial, crowd-sourced data

We describe an architecture for a decentralised data market for applications in which agents are incentivised to collaborate to crowd-source their data. The architecture is designed to reward data that furthers the market's collective goal, and distributes reward fairly to all those that contribute with their data. We show that the architecture is resilient to Sybil, wormhole, and data poisoning attacks. In order to evaluate the resilience of the architecture, we characterise its breakdown points for various adversarial threat models in an automotive use case.

cs.DS↗

arXiv2022

A DLT enabled smart mask system to enable social compliance

As Covid-19 remains a cause of concern, especially due to its mutations, wearing masks correctly and efficiently remains a priority in order to limit the spread of the disease. In this paper we present a wearable smart-mask prototype using concepts from Internet of Things, Control Theory and Distributed Ledger Technologies. Its purpose is to encourage people to comply with social distancing norms, through the use of incentives. The smart mask is designed to monitor Carbon Dioxide and Total Volatile Organic Compounds concentrations. The detected data is appended to a DAG-based DLT, named the IOTA Tangle. The IOTA Tangle ensures that the data is secure and immutable and acts as a communication backbone for the incentive mechanism. A hardware-in-the-loop simulation, based on indoor positioning, is developed to validate the effectiveness of the designed prototype.

cs.CY↗

arXiv2022

Feedback control for distributed ledgers: An attack mitigation policy for DAG-based DLTs

In this paper we present a feedback approach to the design of an attack mitigation policy for DAG-based Distributed Ledgers. We develop a model to analyse the behaviour of the ledger under the so called Tips Inflation Attack and we design a control strategy to counteract this attack strategy. The efficacy of this approach is showcased through a theoretical analysis, in the form of two theorems about the stability properties of the ledger with and without the controller, and extensive Monte Carlo simulations of an agent-based model of the distributed ledger.

eess.SY↗

arXiv2021

Secure Access Control for DAG-based Distributed Ledgers

Access control is a fundamental component of the design of distributed ledgers, influencing many aspects of their design, such as fairness, efficiency, traditional notions of network security, and adversarial attacks such as Denial-of-Service (DoS) attacks. In this work, we consider the security of a recently proposed access control protocol for Directed Acyclic Graph-based distributed ledgers. We present a number of attack scenarios and potential vulnerabilities of the protocol and introduce a number of additional features which enhance its resilience. Specifically, a blacklisting algorithm, which is based on a reputation-weighted threshold, is introduced to handle both spamming and multi-rate malicious attackers. The introduction of a solidification request component is also introduced to ensure the fairness and consistency of network in the presence of attacks. Finally, a timestamp component is also introduced to maintain the consistency of the network in the presence of multi-rate attackers. Simulations to illustrate the efficacy and robustness of the revised protocol are also described.

cs.CR↗

arXiv2021

On node ranking in graphs

The ranking of nodes in a network according to their ``importance'' is a classic problem that has attracted the interest of different scientific communities in the last decades. The current COVID-19 pandemic has recently rejuvenated the interest in this problem, as it is related to the selection of which individuals should be tested in a population of asymptomatic individuals, or which individuals should be vaccinated first. Motivated by the COVID-19 spreading dynamics, in this paper we review the most popular methods for node ranking in undirected unweighted graphs, and compare their performance in a benchmark realistic network, that takes into account the community-based structure of society. Also, we generalize a classic benchmark network originally proposed by Newman for ranking nodes in unweighted graphs, to show how ranks change in the weighted case.

physics.soc-ph↗

arXiv2021

Personalised Feedback Control, Social Contracts, and Compliance Strategies for Ensembles

This paper describes the use of Distributed Ledger Technologies as a mean to enforce social contracts and to orchestrate the behaviour of agents in a smart city environment. Specifically, we present a scheme to price personalised risk in sharing economy applications. We provide proofs for the convergence of the proposed stochastic system and we validate our approach through the use of extensive Monte Carlo simulations.

math.OC↗