graph neural networks

基于图神经网络的程序代码语义理解与重构

This paper proposes a novel approach to program code semantic understanding and reconstruction leveraging Graph Neural Networks (GNNs). The core idea is to directly learn semantic relationships within program code by representing it as a graph structure. Nodes in the graph represent code elements such as functions, variables, and statements, while edges represent the semantic relationships between them, such as call relationships and dependencies. The GNN learns node and edge features from this graph structure, enabling automated code understanding, reconstruction, and optimization. Traditional methods relying on manual analysis and rule-based engines suffer from low efficiency and potential errors. This work demonstrates the effectiveness of GNNs in automating these tasks, offering a more scalable and robust solution for code manipulation. The presented framework aims to improve code maintainability, reduce development time, and facilitate code optimization through a deeper understanding of code semantics. The key contribution lies in the application of GNNs to the specific domain of program code, moving beyond traditional symbolic representations and enabling a more nuanced and effective approach to code analysis and transformation.

基于图的自适应图神经网络

This paper introduces a novel graph neural network (GNN) architecture, termed "Adaptive Graph Neural Network" (AGNN), designed to dynamically adjust its graph structure in response to data fluctuations. The AGNN's core mechanism centers around a dynamic parameterization of the graph, enabling efficient knowledge transfer and generalization. We demonstrate the AGNN's effectiveness through a series of experiments on a diverse dataset, showcasing its superior performance compared to existing GNN models in scenarios requiring adaptive learning. The paper highlights the benefits of this approach, including improved robustness and generalization capabilities.

基于图的机器学习的动态路径规划

This paper investigates the application of Graph Neural Networks (GNNs) to dynamic path planning, offering a novel approach to address the limitations of traditional methods. Dynamic path planning necessitates the ability to adapt to changing environmental conditions and real-time data, which is often challenging with static maps. This work proposes a GNN-based algorithm that dynamically updates path strategies based on node and edge changes, leading to more intelligent and robust path generation. We explore the benefits of this approach in scenarios involving unpredictable environments and varying demand. The core mechanism leverages GNNs to predict future path trajectories, enabling continuous adaptation and optimization. This research demonstrates the potential of GNNs to revolutionize dynamic path planning, offering improved efficiency and resilience.

GDAN-ID: Subject-independent EEG emotion recognition via graph diffusion attention network with INRDecoder

基于图的复杂性学习的自适应几何建模

This paper introduces a novel approach to geometric modeling that leverages graph neural networks (GNNs) for adaptive complexity learning. Traditional methods often require manual design of complex geometric models, leading to high modeling costs. We propose a system that dynamically adjusts the complexity of GNNs based on the model's computational cost, enabling automatic adaptation and improved model performance. This research explores the use of GNNs to represent and learn complex geometric structures, demonstrating a method for automated complexity control and reduced modeling effort. The core mechanism centers around utilizing GNNs to represent the geometric structure and adjust network parameters based on the model's computational demands. This results in a system capable of generating and maintaining highly complex geometric models with minimal manual intervention.

基于图神经网络的系统状态预测与协同控制

This paper introduces a novel system of framework for predicting system states and enabling collaborative control utilizing graph neural networks (GNNs). The core idea is to leverage graph structure to model system dynamics, identify key nodes and their relationships, and facilitate efficient control strategies. We propose a GNN architecture that learns node representations based on the graph topology, enabling the model to effectively capture dependencies and predict future states. The proposed system demonstrates improved performance compared to traditional methods in predicting system behavior, especially in scenarios with complex dynamics. Furthermore, we showcase the potential for enhanced control through coordinated actions based on the identified graph structure. The research focuses on achieving optimal performance through careful design of the GNN architecture and the selection of appropriate node representation learning techniques. This work contributes to the advancement of intelligent systems by providing a robust and adaptable framework for state prediction and control.

##基于图神经网络的复杂性建模

This paper introduces a novel framework for complex system modeling based on graph neural networks (GNNs). We propose a method to automatically learn complex system structures from data, facilitating efficient complexity analysis. Traditional approaches often require manual model design, while GNNs offer a powerful mechanism for learning intricate structural representations. The proposed framework leverages GNNs to represent the system as a graph, where nodes represent elements and edges represent relationships. The model learns node embeddings that capture the system's overall structure, enabling efficient complexity analysis. We demonstrate the effectiveness of this approach through several case studies, highlighting its ability to capture intricate patterns and provide insightful complexity assessments. This work represents a significant advancement in complex system modeling, offering a more automated and flexible approach compared to existing methods.

Beyond Model Complexity: A Reproducible Comparison of Classical Machine Learning, Matrix Factorization, Graph Embeddings, and LightGCN for Recommendation

Recommender systems increasingly incorporate graph embeddings and graph neural networks to capture high-order relationships between users and items. However, the additional complexity of these approaches does not necessarily guarantee better recommendation quality than strong classical and latent-factor baselines. This study presents a reproducible comparison of six recommendation models representing four methodological families: Logistic Regression and Random Forest; Matrix Factorization with Bayesian Personalized Ranking; DeepWalk and node2vec; and LightGCN. The experiments were conducted on the MovieLens 1M dataset using a per-user temporal split. For each user, the most recent positive interaction was assigned to testing, the preceding interaction to validation, and all earlier positive interactions to training. The primary evaluation used identical candidate sets containing one held-out positive movie and 99 sampled unobserved movies. Performance was measured using Recall, Precision, Hit Rate, and NDCG at multiple cutoffs, complemented by bootstrap confidence intervals, paired statistical tests, computational-efficiency measurements, and analyses by user activity and movie popularity. Matrix Factorization achieved the best overall performance, reaching a Recall@10 of 0.7458 and an NDCG@10 of 0.4558, representing an approximately 56% improvement in NDCG@10 over Random Forest, the strongest classical baseline. Validation-based tuning improved LightGCN to an NDCG@10 of 0.2875; it significantly outperformed Logistic Regression but remained statistically indistinguishable from Random Forest after Holm correction. Tuned node2vec also significantly outperformed DeepWalk, reaching an NDCG@10 of 0.1593, although both random-walk embedding methods’ results remained substantially below than the strongest baselines. Popularity-based analysis further revealed that classical models and LightGCN achieved substantially higher ranking effectiveness for popular movies, whereas Matrix Factorization maintained comparatively stronger performance for less-popular items. These findings show that under the evaluated setting, greater model complexity did not consistently translate into higher recommendation effectiveness, and they thus highlight the importance of strong baselines, model tuning, standardized evaluation, and reproducible experimental protocols.

A privacy-preserving federated spatiotemporal dynamic graph neural network framework for epileptic seizure prediction

Abstract Epilepsy ranks among the most prevalent and debilitating neurological disorders, globally affecting an estimated 50 million individuals across all age groups and socioeconomic backgrounds [WHO, 2019]. In recent years, research on its prediction methods has made significant progress, driven by advancements in artificial intelligence technology. Epilepsy prediction models based on electroencephalogram (EEG) signals and deep learning have become an important research direction in the field of neuroscience, and related research results have shown an exponential growth trend. However, there are still several key problems that need to be solved in existing research: First, the mainstream model architecture are focused on traditional neural network framework, exclusively trained in centralized settings that are incompatible with the privacy regulations and data-sharing constraints governing real-world clinical environments; second, the deep learning approaches, including recent Graph Neural Network (GNN) models, using static graph modeling methods, which ignore the dynamic network topological evolution characteristics of EEG signals in the time-varying process, fails to effectively explore the high-order nonlinear correlation characteristics contained in the topological structure of brain functional networks; To address this limitation, this paper proposes a patient-dependent privacy-preserving Federated Learning framework that integrates an epilepsy prediction model DygonNet based on spatiotemporal dynamic graph neural network, a local learning model at each federated client, deployed within a cloud-based simulation environment. In the proposed architecture, the CHB-MIT, SWEC-ETHZ and the TJU-HH iEEG datasets are treated as three independent federated clients representing distinct clinical sites, each performing federated training on their private EEG data, while a central cloud server simulated on Google Colab Pro aggregates the model updates using the Federated Averaging (FedAvg) and FedProx algorithms without ever accessing raw patient recordings. The model defines the dynamic graph structure of EEG signals, innovatively introduces the Transformer model and dynamic graph neural network into the field of epilepsy prediction to fully learn the temporal and spatial characteristics of EEG signals, and proposes a hierarchical graph pooling mechanism based on the attention mechanism in a Federated environment. Experiments show that the model shows excellent epilepsy prediction performance on both public and private datasets.

Yield-graph: multi-stage growth-aware maize yield prediction via graph neural networks

Activated permutation entropy for graph signals

Abstract Nonlinear dynamics play an important role in the analysis of signals. A popular, readily interpretable nonlinear measure is Permutation Entropy (PE). PE has recently been extended for the analysis of graph signals, thus providing a framework for non-linear analysis of data sampled on irregular domains. Integrating ideas from Graph Signal Processing (GSP) and Machine Learning, we introduce Activated Permutation Entropy ($APE_G$) by combining continuous information present in graph signals with traditional ordinal analysis and introducing an ordinal activation function (OAF) akin to the one of neural networks. We also formally extend ordinal contrasts to the graph domain. Activated versions of ordinal contrasts of length 3 are introduced and their advantage is shown in experiments from various domains. Building on recent work in ordinal analysis, we further demonstrate how we can find the continuous ordinal pattern that maximizes the $APE_G$ of various dynamical systems. Simulations with popular non-linear maps and analysis of real-life MRI data show the validity of $APE_G$ and potential benefits over the state of the art. By extending very recent concepts related to permutation entropy to the graph domain, we expect to accelerate the development of improved graph-based entropy methods that enable nonlinear analysis of broader data types and establishing relationships with emerging ideas in data science.

A Fourier physics-informed graph neural network for Allen–Cahn and Conservative Allen–Cahn equations