Neighbor Nodes Research Articles

Rapid motion planning of manipulator in three-dimensional space under multiple scenes

Most working scenes of industrial robots are static scenes and from a previous static scene to a current static scene is a multi-scene. Finding optimal paths with limited time is difficult for motion planning in a high-dimensional space or in multiple scenes. The low efficiency of motion planning of rapidly-exploring random tree star in high-dimensional spaces, low adaptability of global replanning to multiple scenes are addressed by proposing the local replanning based on goal dynamically-guiding rapidly-exploring random tree star (LR-GD-RRT*). The algorithm contributes to fast path tree exploration and multi-scene motion planning. For path tree exploration, sampling points heuristically generating and new nodes growth by goal dynamically-guiding are proposed to reduce the blind and ineffective searches. Moreover, dynamic adjustment of size of new node neighborhood according to density of the obstacles is proposed to search for more neighbor nodes to optimize the path and also to reduce ineffective computation to improve efficiency. For multi-scene, three steps of trimming, re-exploration, and reconnection for local replanning are proposed. For path tree exploration, simulations in 2D plane, 3D space, and the manipulator show that GD-RRT* improves convergence speed, shortens path length and search time, compared with RRT*. For multi-scene, simulations in 3D space and with the manipulator show that the local replanning of the current scene has both lower path cost and higher planning efficiency compared with the global replanning of the previous scene. Motion of the six-degree-of-freedom robot end in a real scene also verifies the effectiveness of the LR-GD-RRT*.

Aerodynamic Drag Coefficient Prediction of a Spike Blunt Body Based on K-Nearest Neighbors

Spike blunt bodies are a method to reduce drag when a body moves at speeds above sound. Several numerical works based on computational fluid dynamics (CFD) have deeply studied fluid performance and highlighted its advantages. However, most documentation focuses on modifying spike physical properties while keeping constant supersonic or hypersonic flow conditions. In recent years, machine learning models have emerged as viable tools to predict values in almost any field, including aerodynamics. In the case of CFD, many models have been explored, such as support vector regression, ensemble methods, and artificial neural networks. However, a simple and easy-to-implement method such as k-Nearest Neighbors has not been extensively explored. This work extrapoled k-Nearest Neighbors to predict the drag coefficient of a spike blunt body for a range of supersonic and hypersonic speeds based on drag data obtained from CFD analysis. The parametric study of the spike blunt body was performed considering body diameter, spike length, and freestream Mach number as input variables. The algorithm presents proper predictions, with errors less than 5% for the drag coefficient and considering a minimum of three neighbor nodes. The k-NN was compared again Kriging model and k-NN presents a better accuracy. The above validates the flexibility of the method and shows a new area of opportunity for the calculation of aerodynamic properties.

SGNN-T: Space graph neural network coupled transformer for molecular property prediction

Molecular properties play a crucial role in material discovery, protein interaction and drug development. The appearance of Graph Neural Network (GNN) significantly improved the performance of molecular property prediction. However, nodes in GNN only update the features of neighbor nodes, resulting in insufficient ability to encode global feature information. The self- attention mechanism in transformer can encode the global information except for local information of molecules, while its spatial information is insufficient. Since molecules are three-dimensional spatial structures, spatial geometry information is an important attribute for molecules properties. To consider these factors, a network model of Space Graph Neural Network coupled Transformer (SGNN-T) is proposed in this paper which can combine global and local molecule information with three-dimensional spatial structures for molecular properties prediction. In this model, Graph neural network Geometric Feature Fusion Module (GGFF) and Transformer Spatial Geometric Feature Enhancement Module (TSGFE) are included to enhance the spatial geometry learning ability of the network. The GGFF module constructs a parallel graph neural network by thinking over atoms, bonds and bond angles at the same time which effectively complements the spatial information of the network by leading into bond angles than normal GNN. The TSGFE module introduces the coordinates and centrality degree features coupled with the features by GGFF into transformer to further enhance the geometric expression ability of the module. Through these two parts, SGNN-T model can encode local and global information of molecules at the same time. Property prediction experiments are executed on the QM9, OMDB and MEGNet dataset. The results of MAE show the proposed model has the best performance than the popular models.

FLSec-RPL: a fuzzy logic-based intrusion detection scheme for securing RPL-based IoT networks against DIO neighbor suppression attacks

The Internet of Things (IoT) has gained popularity and is widely used in modern society. The growth in the sizes of IoT networks with more internet-connected devices has led to concerns regarding privacy and security. In particular, related to the routing protocol for low-power and lossy networks (RPL), which lacks robust security functions, many IoT devices in RPL networks are resource-constrained, with limited computing power, bandwidth, memory, and battery life. This causes them to face various vulnerabilities and potential attacks, such as DIO neighbor suppression attacks. This type of attack specifically targets neighboring nodes through DIO messages and poses a significant security threat to RPL-based IoT networks. Recent studies have proposed methods for detecting and mitigating this attack; however, they produce high false-positive and false-negative rates in detection tasks and cannot fully protect RPL networks against this attack type. In this paper, we propose a novel fuzzy logic-based intrusion detection scheme to secure the RPL protocol (FLSec-RPL) to protect against this attack. Our method is built of three key phases consecutively: (1) it tracks attack activity variables to determine potential malicious behaviors; (2) it performs fuzzy logic-based intrusion detection to identify malicious neighbor nodes; and (3) it provides a detection validation and blocking mechanism to ensure that both malicious and suspected malicious nodes are accurately detected and blocked. To evaluate the effectiveness of our method, we conduct comprehensive experiments across diverse scenarios, including Static-RPL and Mobile-RPL networks. We compare the performance of our proposed method with that of the state-of-the-art methods. The results demonstrate that our method outperforms existing methods in terms of the detection accuracy, F1 score, power consumption, end-to-end delay, and packet delivery ratio metrics.

Dynamic category-sensitive hypergraph inferring and homo-heterogeneous neighbor feature learning for drug-related microbe prediction.

The microbes in human body play a crucial role in influencing the functions of drugs, as they can regulate the activities and toxicities of drugs. Most recent methods for predicting drug-microbe associations are based on graph learning. However, the relationships among multiple drugs and microbes are complex, diverse, and heterogeneous. Existing methods often fail to fully model the relationships. In addition, the attributes of drug-microbe pairs exhibit long-distance spatial correlations, which previous methods have not integrated effectively. We propose a new prediction method named DHDMP which is designed to encode the relationships among multiple drugs and microbes and integrate the attributes of various neighbor nodes along with the pairwise long-distance correlations. First, we construct a hypergraph with dynamic topology, where each hyperedge represents a specific relationship among multiple drug nodes and microbe nodes. Considering the heterogeneity of node attributes across different categories, we developed a node category-sensitive hypergraph convolution network to encode these diverse relationships. Second, we construct homogeneous graphs for drugs and microbes respectively, as well as drug-microbe heterogeneous graph, facilitating the integration of features from both homogeneous and heterogeneous neighbors of each target node. Third, we introduce a graph convolutional network with cross-graph feature propagation ability to transfer node features from homogeneous to heterogeneous graphs for enhanced neighbor feature representation learning. The propagation strategy aids in the deep fusion of features from both types of neighbors. Finally, we design spatial cross-attention to encode the attributes of drug-microbe pairs, revealing long-distance correlations among multiple pairwise attribute patches. The comprehensive comparison experiments showed our method outperformed state-of-the-art methods for drug-microbe association prediction. The ablation studies demonstrated the effectiveness of node category-sensitive hypergraph convolution network, graph convolutional network with cross-graph feature propagation, and spatial cross-attention. Case studies on three drugs further showed DHDMP's potential application in discovering the reliable candidate microbes for the interested drugs. Source codes and supplementary materials are available at https://github.com/pingxuan-hlju/DHDMP.

NLA-GNN: Non-local information aggregated graph neural network for heterogeneous graph embedding

Heterogeneous graphs are ubiquitous in the real world. Recent methods aim to obtain meaningful low-dimensional node embeddings from heterogeneous graphs to facilitate downstream applications. However, most existing methods tend to consider the local information but ignore the non-local information. This paper proposes a novel Non-Local Information Aggregated Graph Neural Network (NLA-GNN) that aggregate not only the local information from neighbor nodes but also non-local information from distant nodes. Specifically, Local aggregation modules in NLA-GNN utilize the attention mechanism to generate potentially valuable metapaths and use them to aggregate local information. In contrast, non-local aggregation modules adopt a two-step approach, and each step uses an attention-guided method to sort nodes into node sequences and aggregate information with the methods designed for sequential data. Experiment results on three heterogeneous graph datasets demonstrate the performance of NLA-GNN over state-of-the-arts and the necessity of non-local aggregation in heterogeneous graphs.

Possible mechanism of action potential propagation mediated by static electric field: A novel assumption of understanding nerve interaction and ephaptic coupling

The generation and propagation of physical signals in living biosystems are continuous issues. Traditional Hodgkin-Huxley model based on ionic current conduction could not explain the fast transmission of action potential in myelinated axons and factors influencing action potential velocity. We propose that the ion flow induced by Nav channel generates near field quasi-static electric field at extracellular space, termed as an ephaptic field which is able to excite nearby passive axons. Our simulation indicates that the static electric field produced by sodium ion channels in one node of Ranvier is improbable to stimulate the ion channels in the adjacent neighboring node. However, the ion channel ring in one node of Ranvier could induce the shift of membrane potential (0.01 mV) on the node at nearby axons (100 μm) in a bundle of axon synchronously, suggesting zig-zag propagation of action potential. Together with the superposition effect of ephaptic feedback field generated by the synchronized movement of adjacent parallel axons stimulate the adjacent node of the original axon, strengthen the action potential to travel in a zig-zag pattern. Our model also provides an explanation for the rapid velocity of action potential propagation reported in experimental studies.

Byzantine-Resilient Second-Order Consensus in Networked Systems.

This article studies the second-order consensus problem in networked systems containing the so-called Byzantine misbehaving nodes when only an upper bound on either the local or the total number of misbehaving nodes is known. The existing results on this subject are limited to malicious/faulty model of misbehavior. Moreover, existing results consider consensus among normal nodes in only one of the two states, with the other state converging to either zero or a predefined value. In this article, a distributed control algorithm capable of withstanding both locally bounded and totally bounded Byzantine misbehavior is proposed. When employing the proposed algorithm, the normal nodes use a combination of the two relative state values obtained from their neighboring nodes to decide which neighbors should be ignored. By introducing an underlying virtual network, conditions on the robustness of the communication network topology for consensus on both states are established. Numerical simulation results are presented to illustrate the effectiveness of the proposed control algorithm.

An Adaptive 3D Neighbor Discovery and Tracking Algorithm in Battlefield Flying Ad Hoc Networks with Directional Antennas.

Neighbor discovery and tracking with directional antennas in flying ad hoc networks (FANETs) is a challenging issue because of dispersed node distribution and irregular maneuvers in three-dimensional (3D) space. In this paper, we propose an adaptive 3D neighbor discovery and tracking algorithm in battlefield FANETs with directional antennas. With time synchronization, a flying node transmits/receives the neighbor discovery packets sequentially in each beam around it to execute a two-way handshake for neighbor discovery. The transmitting or receiving status of each discovery slot depends on the binary code corresponding to the identification of the node. Discovered neighbor nodes exchange their 3D positions in tracking slots periodically for node tracking, and the maximum tracking period is determined by node velocity, beamwidth, and the minimum distance between nodes. By configuring the relevant parameters, the proposed algorithm can also apply to two-dimensional planar ad hoc networks. The simulation results suggest that the proposed algorithm can achieve shorter neighbor discovery time and longer link survival time in comparison with the random scanning algorithm in scenarios with narrow beamwidth and wide moving area. When the frame length increases, the protocol overhead decreases but the average neighbor discovery time increases. The suitable frame length should be determined based on the network range, node count, beamwidth, and node mobility characteristics.

HCP-PIGN: Efficient heat conduction prediction by physics-informed graph convolutional neural network

This work proposes a novel surrogate model (noted as HCP-PIGN) combining two groups of neural networks: i.e., the physics-informed and the graph convolutional neural networks (noted as PINN and GCN). It aims to tackle the existing challenges: pixelated pre-processing of data and large amounts of training data. For predicting 2D steady-state heat conduction, the GCN acting as the prediction module, considering the interdependence between unstructured and neighboring nodes. The PINN serving as the physical constraint module, embeds governing equations into the neural network’s loss function. The HCP-PIGN model obtains precise predictions with diverse geometries and within milliseconds. The predictive performance of HCP-PIGN was further compared with three network structures: i.e., the physics-informed fully connected neural network (noted as FNN), purely data-driven based FNN, and GCN. The results indicate that HCP-PIGN has the lowest error of temperature field predictions, which are below 3 % and 1.3 % for the max and mean relative errors, respectively. The improvements of 28.1% and 34.6% in accuracy are achieved over the pure data-driven GCN, and the physics-driven FNN, respectively. Therefore, the proposed HCP-PIGN model improves the physical prior knowledge and model’s adaptabilities to geometry variations, resulting in superior performances.

ProcSAGE: an efficient host threat detection method based on graph representation learning

Advanced Persistent Threats (APTs) achieves internal networks penetration through multiple methods, making it difficult to detect attack clues solely through boundary defense measures. To address this challenge, some research has proposed threat detection methods based on provenance graphs, which leverage entity relationships such as processes, files, and sockets found in host audit logs. However, these methods are generally inefficient, especially when faced with massive audit logs and the computational resource-intensive nature of graph algorithms. Effectively and economically extracting APT attack clues from massive system audit logs remains a significant challenge. To tackle this problem, this paper introduces the ProcSAGE method, which detects threats based on abnormal behavior patterns, offering high accuracy, low cost, and independence from expert knowledge. ProcSAGE focuses on processes or threads in host audit logs during the graph construction phase to effectively control the scale of provenance graphs and reduce performance overhead. Additionally, in the feature extraction phase, ProcSAGE considers information about the processes or threads themselves and their neighboring nodes to accurately characterize them and enhance model accuracy. In order to verify the effectiveness of the ProcSAGE method, this study conducted a comprehensive evaluation on the StreamSpot dataset. The experimental results show that the ProcSAGE method can significantly reduce the time and memory consumption in the threat detection process while improving the accuracy, and the optimization effect becomes more significant as the data size expands.

A multi-hierarchical aggregation-based graph convolutional network for industrial knowledge graph embedding towards cognitive intelligent manufacturing

The rapid development and widespread applications of cognitive computing technologies have led to a paradigm shift towards cognitive intelligent development in manufacturing, where knowledge plays an increasingly important role in enabling higher levels of cognition. Knowledge graph (KG) has emerged as one of the essential tools in cognitive intelligent manufacturing and its completion would significantly impact the quality of knowledge. To facilitate effective knowledge management, KG embedding has proven to be an effective approach for KG completion. However, existing models have deficiencies in achieving relation-specific transformations, differentiating the neighbor nodes, and exploiting the intermediate information generated during the KG embedding learning process, which is prone to limit model performance and hinder successful applications. To address these limitations, this paper proposes a new multi-hierarchical aggregation-based graph convolutional network (GCN), consisting of relation-aware, entity-aware, and across-block aggregation. A parallel relation and entity-aware aggregation (PREA) block is established to simultaneously perform relation-specific transformations and entity-differentiated learning. Additionally, an across-block aggregation is constructed to efficiently integrate extracted information from the intermediate stacked block. To demonstrate the effectiveness and superiority of the proposed approach, 3D printing KG is constructed, which is a representative knowledge-intensive industry linking to a variety of aspects like raw materials, adhesion, usages, etc. Extensive experiments are conducted based on the link prediction task. Illustrative examples are provided to reveal the practical implementation of the proposed method, along with technical details and insightful opinions, exhibiting its promising applications.

Cluster-based Graph Collaborative Filtering

Graph Convolution Networks (GCNs) have significantly succeeded in learning user and item representations for recommendation systems. The core of their efficacy is the ability to explicitly exploit the collaborative signals from both the first- and high-order neighboring nodes. However, most existing GCN-based methods overlook the multiple interests of users while performing high-order graph convolution. Thus, the noisy information from unreliable neighbor nodes ( \(e.g.\) , users with dissimilar interests) negatively impacts the representation learning of the target node. Additionally, conducting graph convolution operations without differentiating high-order neighbors suffers the over-smoothing issue when stacking more layers, resulting in performance degradation. In this paper, we aim to capture more valuable information from high-order neighboring nodes while avoiding noise for better representation learning of the target node. To achieve this goal, we propose a novel GCN-based recommendation model, termed Cluster-based Graph Collaborative Filtering (ClusterGCF). This model performs high-order graph convolution on cluster-specific graphs, which are constructed by capturing the multiple interests of users and identifying the common interests among them. Specifically, we design an unsupervised and optimizable soft node clustering approach to classify user and item nodes into multiple clusters. Based on the soft node clustering results and the topology of the user-item interaction graph, we assign the nodes with probabilities for different clusters to construct the cluster-specific graphs. To evaluate the effectiveness of ClusterGCF, we conducted extensive experiments on four publicly available datasets. Experimental results demonstrate that our model can significantly improve recommendation performance.

Predicting potential microbe-disease associations based on dual branch graph convolutional network.

Studying the association between microbes and diseases not only aids in the prevention and diagnosis of diseases, but also provides crucial theoretical support for new drug development and personalized treatment. Due to the time-consuming and costly nature of laboratory-based biological tests to confirm the relationship between microbes and diseases, there is an urgent need for innovative computational frameworks to anticipate new associations between microbes and diseases. Here, we propose a novel computational approach based on a dual branch graph convolutional network (GCN) module, abbreviated as DBGCNMDA, for identifying microbe-disease associations. First, DBGCNMDA calculates the similarity matrix of diseases and microbes by integrating functional similarity and Gaussian association spectrum kernel (GAPK) similarity. Then, semantic information from different biological networks is extracted by two GCN modules from different perspectives. Finally, the scores of microbe-disease associations are predicted based on the extracted features. The main innovation of this method lies in the use of two types of information for microbe/disease similarity assessment. Additionally, we extend the disease nodes to address the issue of insufficient features due to low data dimensionality. We optimize the connectivity between the homogeneous entities using random walk with restart (RWR), and then use the optimized similarity matrix as the initial feature matrix. In terms of network understanding, we design a dual branch GCN module, namely GlobalGCN and LocalGCN, to fine-tune node representations by introducing side information, including homologous neighbour nodes. We evaluate the accuracy of the DBGCNMDA model using five-fold cross-validation (5-fold-CV) technique. The results show that the area under the receiver operating characteristic curve (AUC) and area under the precision versus recall curve (AUPR) of the DBGCNMDA model in the 5-fold-CV are 0.9559 and 0.9630, respectively. The results from the case studies using published experimental data confirm a significant number of predicted associations, indicating that DBGCNMDA is an effective tool for predicting potential microbe-disease associations.

ADPSCAN: Structural Graph Clustering with Adaptive Density Peak Selection and Noise Re-Clustering

Structural graph clustering is a data analysis technique that groups nodes within a graph based on their connectivity and structural similarity. The Structural graph clustering SCAN algorithm, a density-based clustering method, effectively identifies core points and their neighbors within areas of high density to form well-defined clusters. However, the clustering quality of SCAN heavily depends on the input parameters, ϵ and μ, making the clustering results highly sensitive to parameter selection. Different parameter settings can lead to significant differences in clustering results, potentially compromising the accuracy of the clusters. To address this issue, a novel structural graph clustering algorithm based on the adaptive selection of density peaks is proposed in this paper. Unlike traditional methods, our algorithm does not rely on external parameters and eliminates the need for manual selection of density peaks or cluster centers by users. Density peaks are adaptively identified using the generalized extreme value distribution, with consideration of the structural similarities and interdependencies among nodes, and clusters are expanded by incorporating neighboring nodes, enhancing the robustness of the clustering process. Additionally, a distance-based structural similarity method is proposed to re-cluster noise nodes to the correct clusters. Extensive experiments on real and synthetic graph datasets validate the effectiveness of our algorithm. The experiment results show that the ADPSCAN has a superior performance compared with several state-of-the-art (SOTA) graph clustering methods.

Firm Default Prediction by GNN with Gravity-Model Informed Neighbor Node Sampling

AbstractFirm default prediction is important in credit risk management and understanding economic trends. Both practitioners and academic researchers have long studied it. While traditional statistical methods such as discriminant analysis and logistic regression have been used recently, machine learning and deep learning methods have been widely applied. The graph neural network (GNN) is one of the latest applications of deep-learning approaches. With the use of GNNs, it is possible to reflect the non-linear relationships of features among neighboring companies around the target company, whereas ordinary machine learning and deep learning methods focus only on the features of the target company. However, when handling large-scale graphs such as inter-firm networks, it is difficult to apply vanilla GNNs naively. Although uniform neighbor node sampling is commonly used for large-scale graphs, to the best of our knowledge, no research has focused on better sampling methods for GNN applications for default prediction. From the practical viewpoint, it means which companies should be considered with priority for firm default prediction. In this study, we propose a novel gravity model-informed neighbor sampling method based on the estimated transaction volume by utilizing knowledge from econophysics. The scope of this research is to determine whether we can improve default predictions by considering neighboring companies with larger transaction amounts compared to ordinary uniform sampling. We also verified that the proposed method improves the prediction performance and stability compared to GNNs with other sampling techniques and other machine learning methods using real large-scale inter-firm network data.

Intra-Frame Graph Structure and Inter-Frame Bipartite Graph Matching with ReID-Based Occlusion Resilience for Point Cloud Multi-Object Tracking

Three-dimensional multi-object tracking (MOT) using lidar point cloud data is crucial for applications in autonomous driving, smart cities, and robotic navigation. It involves identifying objects in point cloud sequence data and consistently assigning unique identities to them throughout the sequence. Occlusions can lead to missed detections, resulting in incorrect data associations and ID switches. To address these challenges, we propose a novel point cloud multi-object tracker called GBRTracker. Our method integrates an intra-frame graph structure into the backbone to extract and aggregate spatial neighborhood node features, significantly reducing detection misses. We construct an inter-frame bipartite graph for data association and design a sophisticated cost matrix based on the center, box size, velocity, and heading angle. Using a minimum-cost flow algorithm to achieve globally optimal matching, thereby reducing ID switches. For unmatched detections, we design a motion-based re-identification (ReID) feature embedding module, which uses velocity and the heading angle to calculate similarity and association probability, reconnecting them with their corresponding trajectory IDs or initializing new tracks. Our method maintains high accuracy and reliability, significantly reducing ID switches and trajectory fragmentation, even in challenging scenarios. We validate the effectiveness of GBRTracker through comparative and ablation experiments on the NuScenes and Waymo Open Datasets, demonstrating its superiority over state-of-the-art methods.

Multi-graph attention fusion graph neural network for remaining useful life prediction of rolling bearings

Abstract Graph neural network (GNN) has the proven ability to learn feature representations from graph data, and has been utilized for the tasks of predicting the machinery remaining useful life (RUL). However, existing methods only focus on a single graph structure and cannot integrate the correlation information contained in multi-graph structures. To address these issues, a multi-graph structure GNN prediction method with attention fusion (MGAFGNN) is proposed in this paper for GNN-based bearing RUL prediction. Specifically, a multi-channel graph attention module is designed to effectively learn the similar features of node neighbors from different graph data and capture the multi-scale latent features of nodes through the nonlinear transformation. Furthermore, a multi-graph attention fusion module (MGAFM) is proposed to extract the collaborative features from the interaction graph, thereby fusing the feature embeddings from different graph structures. The fused feature representation is sent to the long short-term memory network to further learn the temporal features and achieve RUL prediction. The experimental results on two bearing datasets demonstrate that MGAFGNN outperforms existing methods in terms of prediction performance by effectively incorporating multi-graph structural information.

Spatial-Temporal Dynamic Hypergraph Information Bottleneck for Brain Network Classification.

Recently, Graph Neural Networks (GNNs) have gained widespread application in automatic brain network classification tasks, owing to their ability to directly capture crucial information in non-Euclidean structures. However, two primary challenges persist in this domain. First, within the realm of clinical neuro-medicine, signals from cerebral regions are inevitably contaminated with noise stemming from physiological or external factors. The construction of brain networks heavily relies on set thresholds and feature information within brain regions, making it susceptible to the incorporation of such noises into the brain topology. Additionally, the static nature of the artificially constructed brain network's adjacent structure restricts real-time changes in brain topology. Second, mainstream GNN-based approaches tend to focus solely on capturing information interactions of nearest neighbor nodes, overlooking high-order topology features. In response to these challenges, we propose an adaptive unsupervised Spatial-Temporal Dynamic Hypergraph Information Bottleneck (ST-DHIB) framework for dynamically optimizing brain networks. Specifically, adopting an information theory perspective, Graph Information Bottleneck (GIB) is employed for purifying graph structure, and dynamically updating the processed input brain signals. From a graph theory standpoint, we utilize the designed Hypergraph Neural Network (HGNN) and Bi-LSTM to capture higher-order spatial-temporal context associations among brain channels. Comprehensive patient-specific and cross-patient experiments have been conducted on two available datasets. The results demonstrate the advancement and generalization of the proposed framework.

High-frequency and low-frequency dual-channel graph attention network

Most existing graph convolution layers use learnable or fixed weights to sum up neighbor features to aggregate neighbor information. Since the attention values are always positive, these graph convolution layers perform as low-pass filters, which may result in their poor performance on heterophilic graphs. In this paper, two graph convolutional layers are proposed, HLGAT and NGAT. NGAT is a convolution network using only negative attention values, which only make the aggregation of high-frequency information of neighbor nodes. HLGAT makes the aggregation of low-frequency and high-frequency information by two channels, respectively, and fuses two outputs by using a learnable way. On node-classification task, both NGAT and HLGAT offer significant performance improvement compared to existing methods. The results clearly show that: (1) High-frequency information of neighborhoods plays a decisive role in heterophilic graphs. (2) The aggregation of low-frequency and high-frequency information of neighbor nodes can significantly improve the performance on heterophilic graphs.