Search

Personalized Federated Learning (PFL) which collaboratively trains a federated model while considering local clients under privacy constraints has attracted much attention. Despite its popularity, it has been observed that existing PFL approaches result in sub-optimal solutions when the joint distribution among local clients diverges. To address this issue, we present Federated Modular Network (FedMN), a novel PFL approach that adaptively selects sub-modules from a module pool to assemble heterogeneous neural architectures for different clients. FedMN adopts a light-weighted routing hypernetwork to model the joint distribution on each client and produce the personalized selection of the module blocks for each client. To reduce the communication burden in existing FL, we develop an efficient way to interact between the clients and the server. We conduct extensive experiments on the real-world test beds and the results show both the effectiveness and efficiency of the proposed FedMN over the baselines.

In this work, we show that the same theoretical tools that successfully explain other hydrides systems under pressure seem to be at odds with the recently claimed conventional room temperature superconductivity of the carbonaceous sulfur hydride. We support our conclusions with I) the absence of a dominant low-enthalpy stoichiometry and crystal structure in the ternary phase diagram. II) Only the thermodynamics of C-doping phases appears to be marginally competing in enthalpy against H$_3$S. III) Accurate results of the transition temperature given by ab initio Migdal-Eliashberg calculations differ by more than 110 K to recently theoretical claims explaining the high-temperature superconductivity in carbonaceous-hydrogen sulfide. A novel mechanism of superconductivity or a breakdown of current theories in this system is possibly behind the disagreement., 6 pages, 3 figures

Intuitively, doping represents one of the most promising avenues for optimization of best prospect superconductors (SC) such as conventional high-pressure SCs with record critical temperatures. However, doping at high pressure (HP) is very challenging, and there is not a proved route to achieve it in a controlled fashion. Aided by computing simulations, we show that it may be plausible to start by alloying primary materials and subsequently incorporate high ratios of hydrogen at moderates pressures ($\approx$1.5 Mbar). Our theoretical results evidence the possibility to tune the electronic structure of LaH$_{10}$, increase the density of states at the Fermi level by doping of various elements and hence change their superconducting properties. We found aluminium to increase the occupation at the Fermi level by more than 30 %. Incorporation of other elements such as Si, Ge, H, Ir, Ca, and others with a varying percentage also play in favour to tune the electronic structure. More importantly, these predictions lie in experimentally attainable doping levels. Also, for the first time, we shed light on how the formation of defects and vacancies influence on the electronic structure of a HP-hydride superconductor. The concepts presented in this work can be extended to other high-pressure, hydrogen-based superconductors such as H$_3$S. Arguably, doping is one of the promising paths to reach room-temperature superconductivity, a Holy grail of condensed matter physics., 6 pages, Supplemental Material not included

The interpolation, prediction, and feature analysis of fine-gained air quality are three important topics in the area of urban air computing. The solutions to these topics can provide extremely useful information to support air pollution control, and consequently generate great societal and technical impacts. Most of the existing work solves the three problems separately by different models. In this paper, we propose a general and effective approach to solve the three problems in one model called the Deep Air Learning (DAL). The main idea of DAL lies in embedding feature selection and semi-supervised learning in different layers of the deep learning network. The proposed approach utilizes the information pertaining to the unlabeled spatio-temporal data to improve the performance of the interpolation and the prediction, and performs feature selection and association analysis to reveal the main relevant features to the variation of the air quality. We evaluate our approach with extensive experiments based on real data sources obtained in Beijing, China. Experiments show that DAL is superior to the peer models from the recent literature when solving the topics of interpolation, prediction, and feature analysis of fine-gained air quality.