Domain-adaptive message passing graph neural network

Shen, Xiao; Pan, Shirui; Choi, Kup‐Sze; Zhou, Xi

doi:10.1016/j.neunet.2023.04.038

Cited by 10 publications

(2 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The following section discusses the assumptions that were taken into consideration when creating the WSNs network model, The nature of the source and Source Nodes (SNs) is static 30 One sink is used for the Cluster Head (CH) data‐collecting process. Nodes in the SNs are classified as advanced, moderate, or normal based on their complexity 31 Sink ought to be a supernode that maintains up‐to‐date information about every SN. Data from SNs is aggregated by CH and sends the information to the sink node. In the suggested method, the inter‐data communication methodology is chosen to carry out the data transfer via CH. Once the battery level drops to zero, then it is realized as a dead node. The cluster‐based single‐hop communication approach for WLAN‐enabled IoT focuses on the creation of deep learning basis optimum path routing in wireless communication networks.…”

Section: Proposed Methodsmentioning

confidence: 99%

Energy efficient data dissemination in wireless sensor network enabled IoT using domain‐adaptive message passing graph neural network

Shiny,

Ram,

et al. 2024

Int J Communication

View full text Add to dashboard Cite

SummaryIn the past few years, restricted wireless sensor networks (WSNs) enabled the Internet of Things (IoT) have attracted significant attention and expansion to enhance service delivery and resource efficiency. Dissemination is a service offered by WSN that uses radio transmission and over‐the‐air programming for updating the deployed sensor nodes through online. The centralized data dissemination methods are replaced by the distributed approaches because they affect the drawbacks of a single point of failure, no scalability, and insecurity. Therefore, an Energy Efficient Protocol for Data Dissemination in Wireless Sensor network‐enabled IoT using Domain‐Adaptive Message Passing Graph Neural Network (EEP‐WSN‐IoT‐DMPGNN) is proposed in this paper. The nodes are formed as clusters utilizing the Deep Fuzzy Curriculum Clustering (DFCC) technique that rewards nodes belonging to a given cluster. By using the Crayfish Optimization Algorithm (COA), the Cluster Head (CH) selection optimally chose the ideal CH and satisfies the multiple objective functions, such as energy, delay, traffic density, and distance. Afterward, domain‐adaptive Message Passing Graph Neural Network (DMPGNN) based routing protocol is developed, the input given to the routing protocol includes a sink, action history, future node, and maximum‐distance node, which attains enhanced data transfer in the chosen path. The proposed technique attains a lower no. of dead nodes, lower energy consumption, and higher Network Lifetime while analyzed with existing techniques, such as routing technique depending on deep learning for effectual data transmission in 5G WSN communication (DL‐RPDT‐WSN), Reinforcement‐Learning base energy effectual optimized routing protocol in WSN (RL‐EERP‐WSN), and Energy‐efficient intellectual routing method for IoT‐enabled WSN (EIR‐IoT‐WSN), respectively.

show abstract

Section: Proposed Methodsmentioning

confidence: 99%

Energy efficient data dissemination in wireless sensor network enabled IoT using domain‐adaptive message passing graph neural network

Shiny,

Ram,

et al. 2024

Int J Communication

View full text Add to dashboard Cite

show abstract

“…In recent years, Graph Neural Networks (GNNs) have gained increasing attention in both academia and industry due to their superior performance on numerous web applications, such as classification on web services and pages [15,45], image search [1], web spam detection [2], e-commerce recommendations [13,39,42], and social analysis [24,30,31]. Various GNN models have been developed [3,4,8,14,22,23,35,41,46,47] accordingly. Among them, semi-supervised classification is one of the most extensively studied problems due to the scarce labeled data in real-world applications [12,20,34].…”

Section: Introductionmentioning

confidence: 99%

Node-wise Diffusion for Scalable Graph Learning

Huang

Tang

Liu

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

Graph Neural Networks (GNNs) have shown superior performance for semi-supervised learning of numerous web applications, such as classification on web services and pages, analysis of online social networks, and recommendation in e-commerce. The state of the art derives representations for all nodes in graphs following the same diffusion (message passing) model without discriminating their uniqueness. However, (i) labeled nodes involved in model training usually account for a small portion of graphs in the semisupervised setting, and (ii) different nodes locate at different graph local contexts and it inevitably degrades the representation qualities if treating them undistinguishedly in diffusion.To address the above issues, we develop NDM, a universal nodewise diffusion model, to capture the unique characteristics of each node in diffusion, by which NDM is able to yield high-quality node representations. In what follows, we customize NDM for semisupervised learning and design the NIGCN model. In particular, NIGCN advances the efficiency significantly since it (i) produces representations for labeled nodes only and (ii) adopts well-designed neighbor sampling techniques tailored for node representation generation. Extensive experimental results on various types of web datasets, including citation, social and co-purchasing graphs, not only verify the state-of-the-art effectiveness of NIGCN but also strongly support the remarkable scalability of NIGCN. In particular, NIGCN completes representation generation and training within 10 seconds on the dataset with hundreds of millions of nodes and billions of edges, up to orders of magnitude speedups over the baselines, while achieving the highest F1-scores on classification. CCS CONCEPTS• Computing methodologies → Semi-supervised learning; Neural networks.

show abstract