Search

Hierarchical federated learning (HFL) is a promising distributed deep learning model training paradigm, but it has crucial security concerns arising from adversarial attacks. This research investigates and assesses the security of HFL using a novel methodology by focusing on its resilience against adversarial attacks inference-time and training-time. Through a series of extensive experiments across diverse datasets and attack scenarios, we uncover that HFL demonstrates robustness against untargeted training-time attacks due to its hierarchical structure. However, targeted attacks, particularly backdoor attacks, exploit this architecture, especially when malicious clients are positioned in the overlapping coverage areas of edge servers. Consequently, HFL shows a dual nature in its resilience, showcasing its capability to recover from attacks thanks to its hierarchical aggregation that strengthens its suitability for adversarial training, thereby reinforcing its resistance against inference-time attacks. These insights underscore the necessity for balanced security strategies in HFL systems, leveraging their inherent strengths while effectively mitigating vulnerabilities.

We report the design and operation of a versatile tunnel diode oscillator for high sensitivity (50 ppb) measurements of the magnetic susceptibility of samples over a wide temperature range (1.7-100 K) that can be used with a simple liquid helium storage dewar. The design allows for the application of an electric field across the sample to search for magnetoelectric effects., Comment: 3 pages, 3 figures

It is known that iron selenide superconductors exhibit unique characteristics distinct from iron pnicitides, especially in the electron-doped region. However, acomprehensive study of continuous carrier doping ang the corresponding crystal structures of FeSe is still lacking, mainly due to the difficulties in controlling the carrier density in bulk materials. Here, we report the successful synthesis of a new family of bulk Lix(C3H10N2)0.37FeSe, which features a continue superconducting dome harboring Lifshitz transition within the wide range of 0.06~0.68. We demonstrate that with electron-doped, the anion height of FeSe layers deviates lineraly away from the optimized values of pnictides and pressurized FeSe. This feature leads to anew superconducting zone with unique doping dependence of the electronic structures and strong orbital-selective electronic correlation. Optimal superconductivity is achieved when the Fe 3d t2g orbitals have almost the same intermediate electronic correlation strength, with moderate mass enhancement between 3~4 in the two separate superconducting zone. Our result shed light on archieving unified mechanism of superconductivity in iron-based materials.

Despite their well-known limitations, Reynolds-Averaged Navier-Stokes (RANS) models are still the workhorse tools for turbulent flow simulations in today's engineering application. For many practical flows, the turbulence models are by far the largest source of uncertainty. In this work we develop an open-box, physics-informed Bayesian framework for quantifying model-form uncertainties in RANS simulations. Uncertainties are introduced directly to the Reynolds stresses and are represented with compact parameterization accounting for empirical prior knowledge and physical constraints (e.g., realizability, smoothness, and symmetry). An iterative ensemble Kalman method is used to assimilate the prior knowledge and observation data in a Bayesian framework, and to propagate them to posterior distributions of velocities and other Quantities of Interest (QoIs). We use two representative cases, the flow over periodic hills and the flow in a square duct, to evaluate the performance of the proposed framework. Simulation results suggest that, even with very sparse observations, the posterior mean velocities and other QoIs have significantly better agreement with the benchmark data compared to the baseline results. At most locations the posterior distribution adequately captures the true model error within the developed model form uncertainty bounds. The framework is a major improvement over existing black-box, physics-neutral methods for model-form uncertainty quantification, where prior knowledge and details of the models are not exploited. This approach has potential implications in many fields in which the governing equations are well understood but the model uncertainty comes from unresolved physical processes., Comment: 53 pages, 15 figures

To cope with the growing demand for wireless data and to extend service coverage, future 5G networks will increasingly rely on the use of low powered nodes to support massive connectivity in diverse set of applications and services [1]. To this end, virtualized and mass-scale cloud architectures are proposed as promising technologies for 5G in which all the nodes are connected via a backhaul network and managed centrally by such cloud centers. The significant computing power made available by the cloud technologies has enabled the implementation of sophisticated signal processing algorithms, especially by way of parallel processing, for both interference management and network provision. The latter two are among the major signal processing tasks for 5G due to increased level of frequency sharing, node density, interference and network congestion. This article outlines several theoretical and practical aspects of joint interference management and network provisioning for future 5G networks. A cross-layer optimization framework is proposed for joint user admission, user-base station association, power control, user grouping, transceiver design as well as routing and flow control. We show that many of these cross-layer tasks can be treated in a unified way and implemented in a parallel manner using an efficient algorithmic framework called WMMSE (Weighted MMSE). Some recent developments in this area are highlighted and future research directions are identified.