Data Reconstruction Attacks (DRA) pose a significant threat to Federated Learning (FL) systems by enabling adversaries to infer sensitive training data from local clients. Despite extensive research, the question of how to characterize and assess the risk of DRAs in FL systems remains unresolved due to the lack of a theoretically-grounded risk quantification framework. In this work, we address this gap by introducing Invertibility Loss (InvLoss) to quantify the maximum achievable effectiveness of DRAs for a given data instance and FL model. We derive a tight and computable upper bound for InvLoss and explore its implications from three perspectives. First, we show that DRA risk is governed by the spectral properties of the Jacobian matrix of exchanged model updates or feature embeddings, providing a unified explanation for the effectiveness of defense methods. Second, we develop InvRE, an InvLoss-based DRA risk estimator that offers attack method-agnostic, comprehensive risk evaluation across data instances and model architectures. Third, we propose two adaptive noise perturbation defenses that enhance FL privacy without harming classification accuracy. Extensive experiments on real-world datasets validate our framework, demonstrating its potential for systematic DRA risk evaluation and mitigation in FL systems.
Networked cybernetic and physical systems of the Internet of Things (IoT) immerse civilian and industrial infrastructures into an interconnected and dynamic web of hybrid and mobile devices. The key feature of such systems is the hybrid and tight coupling of mobile and pervasive discrete communications in a continuously evolving environment (discrete computations with predominant continuous dynamics). In the aim of ensuring the correctness and reliability of such heterogeneous infrastructures, we introduce the hybrid π-calculus (HpC), to formally capture both mobility, pervasiveness and hybridisation in infrastructures where the network topology and its communicating entities evolve continuously in the physical world. The π-calculus proposed by Robin Milner et al. is a process calculus that can model mobile communications and computations in a very elegant manner. The HpC we propose is a conservative extension of the classical π-calculus, i.e., the extension is “minimal”, and yet describes mobility, time and physics of systems, while allowing to lift all theoretical results (e.g. bisimulation) to the context of that extension. We showcase the HpC by considering a realistic handover protocol among mobile devices.
We present a data-driven modal-decomposition method that extracts the part of an incoherent internal tidal wave that correlates with the proper orthogonal decomposition (POD) of a turbulent mesoscale flow. This method exploits the a priori knowledge that the incoherent internal tide arises from interactions between an incident wave and the turbulent flow and also exploits the corresponding statistical correlation between the two types of motions. The method is presented and tested in an idealised framework based on the rotating-shallow-water model, where we provide a physical interpretation for the decomposition method based on theoretical considerations. Using idealised simulations with a plane wave propagating through a zonal turbulent jet, we first propose the use of the modal-decomposition method as a data analysis technique to understand how the wave is scattered by the flow. In a second step, we construct an estimation algorithm capable of separating the entangled contributions of the wave and mesoscale motions from a single sea surface height snapshot. This algorithm, which consists of estimating the POD coefficients of the turbulent flow shared by the wave and jet modes, is particularly suitable for configurations where the jet contribution to the sea surface height (SSH) is larger than that of the wave.
