Centralized Method Research Articles

Strengthening stability with centralized event-triggered control system with the disturbances and artificial time delay in wireless connected vehicle platooning (CVSs)

This paper addresses the difficulties with connected vehicle systems (CVSs), particularly with vehicle platooning, are examined in this paper. For leader and follower-connected vehicles, the control protocol (which includes artificial delays, disturbances and proportional gains) is implemented. With tracking error systems, system dynamics are modelled while taking outside influences into consideration. Using creative thinking, a centralized event-triggered control system is implemented to maximize fleet wide communication updates. System stability is guaranteed by this centralized method in combination with quadratic form Lyapunov stability analysis. The risk of zeno behaviour is reduced by an event-triggered communication condition that is activated when a threshold is exceeded. The effectiveness of the centralized event-triggered system in improving stability and resilience in connected vehicle platooning scenarios is evaluated numerically through simulations.

A generalized eigenvector centrality for multilayer networks with inter-layer constraints on adjacent node importance

We present a novel approach for computing a variant of eigenvector centrality for multilayer networks with inter-layer constraints on node importance. Specifically, we consider a multilayer network defined by multiple edge-weighted, potentially directed, graphs over the same set of nodes with each graph representing one layer of the network and no inter-layer edges. As in the standard eigenvector centrality construction, the importance of each node in a given layer is based on the weighted sum of the importance of adjacent nodes in that same layer. Unlike standard eigenvector centrality, we assume that the adjacency relationship and the importance of adjacent nodes may be based on distinct layers. Importantly, this type of centrality constraint is only partially supported by existing frameworks for multilayer eigenvector centrality that use edges between nodes in different layers to capture inter-layer dependencies. For our model, constrained, layer-specific eigenvector centrality values are defined by a system of independent eigenvalue problems and dependent pseudo-eigenvalue problems, whose solution can be efficiently realized using an interleaved power iteration algorithm. We refer to this model, and the associated algorithm, as the Constrained Multilayer Centrality (CMLC) method. The characteristics of this approach, and of standard techniques based on inter-layer edges, are demonstrated on both a simple multilayer network and on a range of random graph models. An R package implementing the CMLC method along with example vignettes is available at https://hrfrost.host.dartmouth.edu/CMLC/.

Reinforcement learning-based SDN routing scheme empowered by causality detection and GNN.

In recent years, with the rapid development of network applications and the increasing demand for high-quality network service, quality-of-service (QoS) routing has emerged as a critical network technology. The application of machine learning techniques, particularly reinforcement learning and graph neural network, has garnered significant attention in addressing this problem. However, existing reinforcement learning methods lack research on the causal impact of agent actions on the interactive environment, and graph neural network fail to effectively represent link features, which are pivotal for routing optimization. Therefore, this study quantifies the causal influence between the intelligent agent and the interactive environment based on causal inference techniques, aiming to guide the intelligent agent in improving the efficiency of exploring the action space. Simultaneously, graph neural network is employed to embed node and link features, and a reward function is designed that comprehensively considers network performance metrics and causality relevance. A centralized reinforcement learning method is proposed to effectively achieve QoS-aware routing in Software-Defined Networking (SDN). Finally, experiments are conducted in a network simulation environment, and metrics such as packet loss, delay, and throughput all outperform the baseline.

Research on Disposal of Non-Performing Assets of Listed Companies

China's economic development is affected by adverse factors such as cyclical and structural adjustment of the domestic economy, but also by the sudden impact of the international political and economic environment, and the complex economic environment at home and abroad has greatly challenged the development of Chinese enterprises. The problem of non-performing assets is a prominent problem facing the high-quality development of China's economy. It requires both enterprises to regulate their own risks and asset management companies to properly and reasonably dispose of non-performing assets to prevent these assets from having an impact on systemic risks and then affecting other economies. Solving the problem of non-performing assets in a reasonable way can effectively deal with the non-performing assets, and then promote and prevent the further spread of the non-performing assets crisis of enterprises. This paper analyzes that the asset disposal mode of "stock plus bond" provides a reference for other NPL disposal methods, and helps other enterprises to dispose NPL efficiently. When dealing with non-performing assets, there are risks from debtors and creditors. To prevent and address these problems, detailed and precise asset-liability analysis, a well-developed restructuring plan, and enhanced collaboration between departments are required." The "equity plus bond" model combines the centralized disposal method of non-performing assets, which can effectively deal with these non-performing assets, better reduce risks and improve operations.

Master–Slave Game Optimal Scheduling for Multi-Agent Integrated Energy System Based on Uncertainty and Demand Response

With the transformation of the energy market from the traditional vertical integrated structure to the interactive competitive structure, the traditional centralized optimization method makes it difficult to reveal the interactive behavior of multi-agent integrated energy systems (MAIES). In this paper, a master–slave game optimal scheduling strategy of MAIES is proposed based on the integrated demand response. Firstly, a master–slave game framework of MAIES is established with an energy management agent as leader, an energy operation agent, an energy storage agent, and a user aggregation agent as followers. Secondly, in view of the wind and solar uncertainty, the Monte Carlo method is used to generate random scenarios, and the k-means clustering method and pre-generation elimination technology are used for scenario reduction. Then, according to different flexible characteristics of loads, a multi-load and multi-type integrated demand response model including electric, thermal, and cold energy is built to fully utilize the regulation role of flexible resources. On this basis, the transaction decision-making models of each agent are constructed, and the existence and uniqueness of the Stackelberg equilibrium solution are proved. Finally, the case simulations demonstrate the effectiveness of the proposed optimal scheduling strategy of MAIES. Compared to the scenario without considering the wind and solar uncertainty and the integrated demand response, the rate of renewable energy curtailment was reduced by 6.03% and the carbon emissions of the system were reduced by 1335.22 kg in the scenario considering the proposed method in this paper.

Predicting Heart Diseases in IoT-Based Electronic Health Records: A Federated Learning Approach

Predicting heart diseases is important for finding them early and treating them effectively. We present a shared learning method for predicting heart diseases using IoT-based electronic health records (EHRs) in this work. Federated learning lets many autonomous IoT devices work together to train a model, while protecting the safety and security of the data. Proposed method uses the fact that IoT devices are spread out to train a global model for predicting heart disease without putting private EHR data in one place. With the EHR data, each IoT device learns a model locally and only sends model changes to a central computer. The computer takes all of these changes and improves the world model. This model is then sent back to the IoT devices to be improved even more. This looping process makes sure that the world model keeps getting better while keeping data private. The proposed method tested by using a real-world collection of EHRs from IoT devices in trials. We looked at how well our shared learning method worked compared to more standard centralized learning methods. Our results show that the pooled learning method makes predictions that are as good as or better than the other methods while protecting data privacy. It also looked at how different IoT device properties, like the amount of data they send and receive and their processing power, affect the shared learning process. It is discovered that devices with more processing power and more data add more to the improvement of the global model. This shows how important it is to choose the right devices in shared learning systems. The paper study shows that pooled learning can be used to predict heart diseases in IoT-based EHRs and that it works well. Our method uses the ability of IoT devices to work together to make accurate predictions while protecting data privacy. This makes it suitable for use in real-life healthcare situations.

Update of high voltage isolation control and monitoring system for HVE-400 ion implanter

HVE-400 ion implanter is special ion implantation equipment for semiconductor materials boron and phosphorus doping. The ion source and extraction deflection system are at high voltage platform, while the corresponding control system is at ground voltage position. The control signals and measurement signals of various parameters at the high-voltage end need to be transmitted between ground voltage and high voltage through optical fibers to isolate high voltage. Upgrading is carried out due to the aging of the optical fiber transmission control and monitoring system, which cannot work stably. The transformation replaces the original distributed single-point control method with an advanced distributed centralized control method, and integrates all control and monitoring functions into an industrial control computer for digital operation and display. In the computer software, two kinds of automatic calculation of ion mass number are designed. After upgrading, the implanter high-voltage platform control and monitoring system features digitalization, centralized control, high reliability, strong anti-interference, fast communication speed, and easy operation.

Identification of top influence users in disseminating information on the 2024 Indonesian National Election

Social media has a vital role in general elections in Indonesia because social media is one of the platforms used by presidential candidates for campaigns to gain public support. General elections in Indonesia occur every five years. Many tweets talk about presidential candidates approaching the national election period. Not least, some buzzers deliberately use Twitter to carry out propaganda against a candidate or to bring down other presidential candidates with their opinions because information can spread widely and quickly on Twitter. Based on this, it is necessary to identify influential users in disseminating information related to the 2024 National Election, especially on Twitter. Various centrality methods were used in this study to identify influence users in sharing information about the 2024 National Election such us Degree Centrality, Closeness Centrality, Harmonic Centrality, Eigenvector Centrality, and Load Centrality. For the evaluation in this study, the results of each method were compared to one another to measure the similarity and correlation between the ranking lists of users who were influential in disseminating information about the 2024 National Election.

Predicting Heart Diseases in IoT-Based Electronic Health Records: A Federated Learning Approach

Predicting heart diseases is important for finding them early and treating them effectively. We present a shared learning method for predicting heart diseases using IoT-based electronic health records (EHRs) in this work. Federated learning lets many autonomous IoT devices work together to train a model, while protecting the safety and security of the data. Proposed method uses the fact that IoT devices are spread out to train a global model for predicting heart disease without putting private EHR data in one place. With the EHR data, each IoT device learns a model locally and only sends model changes to a central computer. The computer takes all of these changes and improves the world model. This model is then sent back to the IoT devices to be improved even more. This looping process makes sure that the world model keeps getting better while keeping data private. The proposed method tested by using a real-world collection of EHRs from IoT devices in trials. We looked at how well our shared learning method worked compared to more standard centralized learning methods. Our results show that the pooled learning method makes predictions that are as good as or better than the other methods while protecting data privacy. It also looked at how different IoT device properties, like the amount of data they send and receive and their processing power, affect the shared learning process. It is discovered that devices with more processing power and more data add more to the improvement of the global model. This shows how important it is to choose the right devices in shared learning systems. The paper study shows that pooled learning can be used to predict heart diseases in IoT-based EHRs and that it works well. Our method uses the ability of IoT devices to work together to make accurate predictions while protecting data privacy. This makes it suitable for use in real-life healthcare situations.

Collaborative inference for treatment effect with distributed data-sharing management in multicenter studies.

Data sharing barriers present paramount challenges arising from multicenter clinical studies where multiple data sources are stored and managed in a distributed fashion at different local study sites. Merging such data sources into a common data storage for a centralized statistical analysis requires a data use agreement, which is often time-consuming. Data merging may become more burdensome when propensity score modeling is involved in the analysis because combining many confounding variables, and systematic incorporation of this additional modeling in a meta-analysis has not been thoroughly investigated in the literature. Motivated from a multicenter clinical trial of basal insulin treatment for reducing the risk of post-transplantation diabetes mellitus, we propose a new inference framework that avoids the merging of subject-level raw data from multiple sites at a centralized facility but needs only the sharing of summary statistics. Unlike the architecture of federated learning, the proposed collaborative inference does not need a center site to combine local results and thus enjoys maximal protection of data privacy and minimal sensitivity to unbalanced data distributions across data sources. We show theoretically and numerically that the new distributed inference approach has little loss of statistical power compared to the centralized method that requires merging the entire data. We present large-sample properties and algorithms for the proposed method. We illustrate its performance by simulation experiments and the motivating example on the differential average treatment effect of basal insulin to lower risk of diabetes among kidney-transplant patients compared to the standard-of-care.

Exploring the potential of federated learning for diffusion model: Training and fine-tuning

Diffusion models, a state-of-the-art generative model, have drawn attention for their capacity to produce high-quality, divers, and flexible content. However, the training of these models typically necessitates large datasets, a task that can be hindered by challenges related to privacy concerns and data distribution constraints. Due to the amount of data and hardware required for large model training, all centralized training will be done by large companies or labs with computing power. Federated Learning provides a decentralized method that allows for model training across several data sources while maintaining the data's localization, reducing privacy threats. This research proposes and evaluate a novel approach for utilizing Federated Learning in the context of diffusion models. This paper investigates the feasibility of training and fine-tuning diffusion models in a federated setting, considering various data distributions and privacy constraints. This study used the Federated Averaging (FedAvg) technique to train the unconditional diffusion model as well as to fine-tune the pre-trained diffusion mode. The experimental results demonstrate that federated training of diffusion models can achieve comparable performance to centralized training methods while preserving data locality. Additionally, Federated Learning can be effectively applied to fine-tune pre-trained diffusion model, enabling adaptation to specific tasks without exposing sensitive data. Overall, this work demonstrates Federated Learning's potential as a useful tool for training and fine-tuning diffusion models in a privacy-preserving manner.

FedDAT: An Approach for Foundation Model Finetuning in Multi-Modal Heterogeneous Federated Learning

Recently, foundation models have exhibited remarkable advancements in multi-modal learning. These models, equipped with millions (or billions) of parameters, typically require a substantial amount of data for finetuning. However, collecting and centralizing training data from diverse sectors becomes challenging due to distinct privacy regulations. Federated Learning (FL) emerges as a promising solution, enabling multiple clients to collaboratively train neural networks without centralizing their local data. To alleviate client computation burdens and communication overheads, previous works have adapted Parameter-efficient Finetuning (PEFT) methods for FL. Hereby, only a small fraction of the model parameters are optimized and communicated during federated communications. Nevertheless, most previous works have focused on a single modality and neglected one common phenomenon, i.e., the presence of data heterogeneity across the clients. Therefore, in this work, we propose a finetuning framework tailored to heterogeneous multi-modal FL, called Federated Dual-Aadapter Teacher (FedDAT). Specifically, our approach leverages a Dual-Adapter Teacher (DAT) to address data heterogeneity by regularizing the client local updates and applying Mutual Knowledge Distillation (MKD) for an efficient knowledge transfer. FedDAT is the first approach that enables an efficient distributed finetuning of foundation models for a variety of heterogeneous Vision-Language tasks. To demonstrate its effectiveness, we conduct extensive experiments on four multi-modality FL benchmarks with different types of data heterogeneity, where FedDAT substantially outperforms the existing centralized PEFT methods adapted for FL.

Bidirectional Contrastive Split Learning for Visual Question Answering

Visual Question Answering (VQA) based on multi-modal data facilitates real-life applications such as home robots and medical diagnoses. One significant challenge is to devise a robust decentralized learning framework for various client models where centralized data collection is refrained due to confidentiality concerns. This work aims to tackle privacy-preserving VQA by decoupling a multi-modal model into representation modules and a contrastive module, leveraging inter-module gradients sharing and inter-client weight sharing. To this end, we propose Bidirectional Contrastive Split Learning (BiCSL) to train a global multi-modal model on the entire data distribution of decentralized clients. We employ the contrastive loss that enables a more efficient self-supervised learning of decentralized modules. Comprehensive experiments are conducted on the VQA-v2 dataset based on five SOTA VQA models, demonstrating the effectiveness of the proposed method. Furthermore, we inspect BiCSL's robustness against a dual-key backdoor attack on VQA. Consequently, BiCSL shows significantly enhanced resilience when exposed to the multi-modal adversarial attack compared to the centralized learning method, which provides a promising approach to decentralized multi-modal learning.

Alterations in degree centrality and functional connectivity in tension-type headache: a resting-state fMRI study.

Previous studies have provided evidence of structural and functional changes in the brains of patients with tension-type headache (TTH). However, investigations of functional connectivity alterations in TTH have been inconclusive. The present study aimed to investigate abnormal intrinsic functional connectivity patterns in patients with TTH through the voxel-wise degree centrality (DC) method as well as functional connectivity (FC) analysis. A total of 33 patients with TTH and 30 healthy controls (HCs) underwent resting-state functional magnetic resonance imaging (rs-fMRI) scanning and were enrolled in the final study. The voxel-wise DC method was performed to quantify abnormalities in the local functional connectivity hubs. Nodes with abnormal DC were used as seeds for further FC analysis to evaluate alterations in functional connectivity patterns. In addition, correlational analyses were performed between abnormal DC and FC values and clinical features. Compared with HCs, patients with TTH had higher DC values in the left middle temporal gyrus (MTG.L) and lower DC values in the left anterior cingulate and paracingulate gyri (ACG.L) (GRF, voxel-wise p < 0.05, cluster-wise p < 0.05, two-tailed). Seed-based FC analyses revealed that patients with TTH showed greater connections between ACG.L and the right cerebellum lobule IX (CR-IX.R), and smaller connections between ACG.L and ACG.L. The MTG.L showed increased FC with the ACG.L, and decreased FC with the right caudate nucleus (CAU.R) and left precuneus (PCUN.L) (GRF, voxel-wise p < 0.05, cluster-wise p < 0.05, two-tailed). Additionally, the DC value of the MTG.L was negatively correlated with the DASS-depression score (p = 0.046, r=-0.350). This preliminary study provides important insights into the pathophysiological mechanisms of TTH.

A timestamp-based log-linear algorithm for solving locally-informed multi-agent finite games

In this paper, the distributed Nash equilibrium seeking problem for multi-agent finite games is considered. Considering the locally-informed communication structure, previous centralized game-theoretic learning dynamic methods are not feasible. To this end, we propose a distributed log-linear algorithm by introducing a timestamp-based communication mechanism that enables agents to acquire estimates of the true action profile from their neighborhoods. It is shown that the proposed distributed log-linear algorithm significantly outperforms the existing best response dynamics regarding both effectiveness and reliability. For illustration, a locally-informed multi-player Sudoku puzzle is introduced, where each cell is taken as a player, and each player fills numbers only based on the local information of its neighboring players. The simulation results validate the effectiveness of our algorithm in solving the locally-informed multi-player Sudoku puzzle. Our methodology provides a viable and benchmark approach to address the fully distributed locally-informed multi-agent joint decision-making problem.

Smart grid optimization considering decentralized power distribution and demand side management

AbstractThe increasing integration of microgrids into distribution networks has highlighted the significance of evaluating and managing intelligent microgrids from both technical and economic perspectives. In this paper, a decentralized approach using agents is employed to optimize the operation of an intelligent microgrid within the telecommunications platform. The decentralized control method comprises two layers. The first layer represents the main microgrid, which includes loads and their controllers, as well as renewable and conventional resources. In the secondary layer, there is a telecommunication platform in which agents can operate as a control processor along with the means of communication. It should be noted that agents interact with the primary layer and neighbouring agents and exchange information with each other. This exchange takes place until the best state of optimization for the power supply occurs. In this study, the operation cost is calculated for decentralized control rules and considering telecommunication links. Also, the effect of performance on cost reduction is examined and compared with normal conditions and centralized methods. It can be seen that the operation cost of the network has decreased to 9.034% after the implementation of the mentioned method in comparison with the normal condition and it has decreased to 6.957% in comparison with the centralized method. Then, using a demand side management program, the cost will be reduced by 2.5%. In the next step, the uncertainty of available resources is taken into account where the uncertainties increase the cost by 7.8%.

A hierarchical federated learning framework for collaborative quality defect inspection in construction

Recent advancements in robotics and deep learning (DL) have made it possible to implement robots in civil infrastructures' quality defect inspection. Robots can reduce human inspectors' workloads and enhance inspection results' reliability by collecting data and automatically identifying quality defects from the raw data. However, current methods for training DL models rely on centralized strategies that require the aggregation of defect data (e.g., uploading to a cloud server), posing concerns about data privacy and security. Thus, this study proposes a three-fold federated learning (FL) framework for training DL models collaboratively, without the need to share local data among construction robots. The framework is specifically applied to image-based crack segmentation, critical for ensuring infrastructures’ safety and serviceability. A lightweight DL model is developed to enable easy implementation on resource-constrained construction robots and to reduce communication costs during federated training. Experimental results show that the proposed FL method outperforms traditional centralized methods. The critical contribution of this study is the hierarchical FL framework, which enables construction robots to leverage big data in a privacy-preserving manner.

Efficient Distributed Hop-Constrained Path Enumeration on Large-Scale Graphs

The enumeration of hop-constrained simple paths is a building block in many graph-based areas. Due to the enormous search spaces in large-scale graphs, a single machine can hardly satisfy the requirements of both efficiency and memory, which causes an urgent need for efficient distributed methods. In practice, it is inevitable to produce plenty of intermediate results when directly extending centralized methods to the distributed environment, thereby causing a memory crisis and weakening the query performance. The state-of-the-art distributed method HybridEnum designed a hybrid search paradigm to enumerate simple paths. However, it makes massive exploration for the redundant vertices not located in any simple path, thereby resulting in poor query performance. To alleviate this problem, we design a distributed approach DistriEnum to optimize query performance and scalability with well-bound memory consumption. Firstly, DistriEnum adopts a graph reduction strategy to rule out the redundant vertices without satisfying the constraint of hop number. Then, a core search paradigm is designed to simultaneously reduce the traversal of shared subpaths and the storage of intermediate results. Moreover, DistriEnum is equipped with a task division strategy to theoretically achieve workload balance. Finally, a vertex migration strategy is devised to reduce the communication cost during the enumeration. The comprehensive experimental results on 10 real-world graphs demonstrate that DistriEnum achieves up to 3 orders of magnitude speedup than HybridEnum in query performance and exhibits superior performances on scalability, communication cost, and memory consumption.

Predicting ride-hailing passenger demand: A POI-based adaptive clustering federated learning approach

Passenger demand prediction is a key task for online ride-hailing platforms to optimize their resource allocation and service quality. However, centralized data collection and mining of massive passengers’ travel data expose serious privacy and security risks. To address this challenge, we propose a POI-based Adaptive Clustering Federated Learning with Spatio-Temporal Graph Attention Gate Recurrent Unit Networks (CFL-STG) algorithm that achieves accurate passenger demand prediction while protecting passenger privacy. Firstly, we propose a localized data storage mechanism for preserving data privacy and facilitating federated learning among different regions. Then, to tackle the statistical heterogeneity of data in federated learning, we introduce a clustering strategy that groups the regions based on their Point of Interest (POI) features, which reflect their demand patterns. For each group, we design a Spatio-Temporal Graph Attention Gate Recurrent Unit Network (STG) as a local model that can capture the spatio-temporal dynamics and inter-regional influence of POI features on passenger demand. Experiments on real data show that the proposed framework outperforms the existing federated learning models and common centralized data mining methods in terms of demand prediction performance and learning efficiency.

Enhancing micro grid sustainability: Genetic-driven reptile search algorithm based carbon emission and cost reduction strategy

Grid architecture is continuously changing in response to modern energy rules that ensure the integration of more renewable sources to lower the carbon footprint. The integration of non-dispatchable renewable energy sources (RESs) using low voltage grids is a possible alternative to a centralized method. Due to a recent trend, microgrids (MGs) are reducing their reliance on the primary grid by incorporating localized sources of electricity including energy storage devices, micro turbines, and fuel cells. Due to the expanding number of spread generating sources with diverse features, power dispatching schedule for distribution micro grids is becoming more challenging. This study offers a dispatch scheduling method from the perspective of an operator. The primary objective of this study is to employ the Genetic-driven Reptile search algorithm (G-RSA) for the dual purpose of minimizing carbon emissions and reducing costs within each microgrid. Additionally, fewer sale and purchase are made from the primary grid. Transmission losses are consequently reduced.