A Federated Graph Neural Network Framework for Privacy-Preserving Personalization

Graph neural network (GNN) is effective in modeling high-order interactions and has been widely used in various personalized applications such as recommendation. However, mainstream personalization methods rely on centralized GNN learning on global graphs, which have considerable privacy risks due to the privacy-sensitive nature of user data. Here, we present a federated GNN framework named FedGNN for both effective and privacy-preserving personalization. Through a privacy-preserving model update method, we can collaboratively train GNN models based on decentralized graphs inferred from local data. To further exploit graph information beyond local interactions, we introduce a privacy-preserving graph expansion protocol to incorporate high-order information under privacy protection. Experimental results on six datasets for personalization in different scenarios show that FedGNN achieves 4.0%~9.6% lower errors than the state-of-the-art federated personalization methods under good privacy protection. FedGNN provides a novel direction to mining decentralized graph data in a privacy-preserving manner for responsible and intelligent personalization.

An Intelligent Diagnosis Method for Machine Fault Based on Federated Learning

In engineering, the fault data unevenly distribute and difficultly share, which causes that the existing fault diagnosis methods cannot recognize the newly added fault types. An intelligent diagnosis method for machine fault is proposed based on federated learning. Firstly, the local fault diagnosis models diagnosing the existing fault data and the newly added fault data are established by deep convolutional neural network. Then, the weight parameters of local models are fused into global model parameters by federated learning. Finally, the global model parameters are transmitted to each local model. Therefore, each local model update into a global shared model which can recognize the newly added fault types. The proposed method is verified by bearing data. Compared with the traditional model, which can only diagnose the existing fault data but cannot recognize newly added fault types, the federated fault diagnosis model fusing weight parameters can diagnose newly added faults without exchanging the data, and the accuracy is 100%. The proposed method provides an effective method to solve the poor sharing of fault data and poor generalization of fault diagnosis model for mechanical equipment.

Robust Probabilistic Discriminative Model Prediction Tracker via Improved Model Update Strategy

In the visual object tracking, the tracking algorithm based on discriminative model prediction have shown favorable performance in recent years. Probabilistic discriminative model prediction (PrDiMP) is a typical tracker based on discriminative model prediction. The PrDiMP evaluates tracking results through output of the tracker to guide online update of the model. However, the tracker output is not always reliable, especially in the case of fast motion, occlusion or background clutter. Simply using the output of the tracker to guide the model update can easily lead to drift. In this paper, we present a robust model update strategy which can effectively integrate maximum response, multi-peaks and detector cues to guide model update of PrDiMP. Furthermore, we have analyzed the impact of different model update strategies on the performance of PrDiMP. Extensive experiments and comparisons with state-of-the-art trackers on the four benchmarks of VOT2018, VOT2019, NFS and OTB100 have proved the effectiveness and advancement of our algorithm.

Enhancing the diagnostic performance of Condition Based Maintenance through the fusion of sensor with maintenance data

Although Maintenance data is crucial for authoritative reporting reasons and is generally used to optimizemaintenance planning in terms of budget, scheduling and logistics, the potentials of the implicit given informationfor Prognostics and Health Management (PHM) frameworks are not yet completely leveraged. Traditional PHMframeworks typically rely only on sensor data to derive a system’s health status, while maintenance, repair andoverhaul (MRO) data is not investigated. However, maintenance data contains valuable information on which partof a system is checked, serviced or replaced. At the same time, maintenance data is necessary for the labelling ofsensor data, the differentiation of multiple failure modes and includes the expert knowledge of the worker. Theoverall goal of the presented work is enable a model update through the integration of this information into atraditional (sensor-based) PHM/condition monitoring framework.In this context, the underlying data bases and structures will be analyzed and a generalized methodology isproposed to include maintenance data directly into the forward-modelling phase of a PHM/condition monitoringframework. The main goal is not only to use the labels derived from maintenance data for evaluation purposes(which is a common practice in PHM research), but to use this data to build a memory of the maintenance andhealth state history and thereby enhance the diagnostic capabilities of the framework. Methods from the field ofProbabilistic Programming and Bayesian Statistics seem promising and are implemented in order to incorporatefor uncertainties and to enable a confidence level for the diagnosis. The proposed concept is developed, tested andassessed in a simulation environment, allowing to investigate the influence of data confidence and label uncertaintyon the results. Furthermore, this allows to derive specific requirements for the input data and hence for the dataacquisition in the real world. The proposed concept is described in a generic way to be applicable on differentengineering domains (e.g. wind turbine or production machinery industry), but it will be tested and evaluated on areal world aviation use case. This concluding use case is defined in the context of the project INDI at TU Darmstadt(Intelligent Data Utilization in Maintenance) in cooperation with the project partner Lufthansa Technik AG.

Local Dynamic Updating Method of Orebody Model Based on Mesh Reconstruction and Mesh Deformation

In this paper, to update the orebody model based on the given interpreted geological information, we present a local dynamic updating method of the orebody model that allows the interactive construction of the constraint deformation conditions and the dynamic updating of the mesh model. The rules for constructing deformation constraints based on the control polylines are discussed. Because only part of the model is updated, the updated mesh is effective and the overall quality is satisfactory. Our main contribution is that we propose a local dynamic updating method for the orebody model based on mesh reconstruction and mesh deformation. This method can automatically update a given 3D orebody model based on a set of unordered geological interpretation lines. Moreover, we implement a deformation neighborhood region search method based on the specified ring radius and a local constrained mesh deformation algorithm for the orebody model. Finally, we test the method and show the model update results with real geological datasets, which proves that this method is effective for the local updating of orebody models.

Research on Synchronization Mechanism of Artificial Intelli-gence Model Based on Belief Measure and Its Application in Pumped Storage Power Station

For the security monitoring of pumped storage power station, a model synchroniza-tion mechanism for cloud edge cooperation framework is proposed. The method uses the belief function to describe the threshold and uses the ping-pong operation strategy to update the model alternately, which solves the problem of artificial intelligence model synchronization and update of edge equipment. The cloud is based on Baidu BML platform, the edge uses customized servers, and the average model update cycle is about three months.

INFRA-BIM FOR BUSINESS PROCESSES’ MANAGEMENT IN ROAD CONSTRUCTION AND OPERATION

Introduction: The article addresses the business processes of the construction and operation of linear road sections. In the course of the study, we analyzed the geographic information system (GIS) of Avtodor company (developer: Indor-Soft, Tomsk) and its use in the operation and maintenance of transport infrastructure facilities at such road sections. In particular, we studied the construction of the Far West Krasnodar Bypass (FWKB), which is part of the North-South transport corridor, and its implementation by means of InfraBIM modeling. Purpose of the study: We aimed to describe the structured GIS database of Avtodor company by connecting various InfraBIM tools for the purposes of data generation and road model update. Methods: Analyzing the GIS of Avtodor company when using an InfraBIM model of road infrastructure, we studied the following: use of GIS features at the facility; asynchronous editing and selection of data without access to the Internet; database update. Results: InfraBIM models of linear road sections, developed in the implementation of such projects as the FWKB, make it possible to optimize time and save money when planning and introducing measures during the operation of transport infrastructure facilities.