Local Neighborhood Information Research Articles

TriClsNet: Surface Reconstruction via Graph-based Triangle Classification

In this paper, we introduce TriClsNet, a novel learning-based network that reconstructs surfaces by reframing the triangle classification problem as a graph node classification problem. An improved graph-based triangle classification module is employed to aggregate information from neighboring triangles, effectively leveraging local neighborhood information and enhancing triangle classification accuracy. Additionally, a self-supervised learning branch is incorporated to predict point cloud normals, aiding our network in better learning local point cloud features. Furthermore, a new loss function is designed to guide our network in effective multi-task learning, encompassing both graph node classification and normal prediction. Comparative experimental results on ShapeNet demonstrate that our method can efficiently perform surface reconstruction, outperforming existing methods in the aspects of preserving surface details, reducing holes, and generalization.

Adaptive local neighborhood information based efficient fuzzy clustering approach

The purpose of clustering is to partition data similar with each other into a same group and partition data dissimilar with each other into different groups. However, in most existing fuzzy clustering approaches, the membership degrees of an individual belonging to different clusters are relied on the different distances between the individual and different cluster centroids, the similarity between the individual and data in different clusters are ignored, besides, the outliers cannot be distinguished effectively. For improving the clustering performance and addressing the problems in most existing fuzzy clustering approaches, based on the concept that data close to each other should be grouped together and be assigned to a same cluster, in this paper, we present a novel efficient fuzzy clustering approach, in which the adaptive local neighborhood information of each data referring to different clusters is taken into consideration; an entirely new idea that the membership degree values of an individual referring to different clusters should not only depend on the distances between the individual and different cluster centers, but also rely on the distances between the individual and several nearest neighbor data in different clusters is put forward; a new scheme to search for the outliers is presented, a method for identifying the different importance of different features is introduced. Experiments on synthetic and publicly real-world datasets were conducted, the clustering results show that the approach put forward can yield consistently satisfying clustering performance and has significant advantages.

PMA-Net: Progressive multi-stage adaptive feature learning for two-view correspondence

Establishing high-quality correspondences is a fundamental step for many computer vision tasks. Leveraging the local consistency of correct correspondences (i.e., inliers) has been a prevalent approach to distinguish them from incorrect correspondences (i.e., outliers). However, the random distribution of a significant proportion of outliers complicates the local neighborhood construction of inliers, making it inevitably contain many outliers, which compromises the reliability of the consistency information. In this paper, we propose a Progressive Multi-Stage Adaptive Feature Learning Network (PMA-Net) to solve this problem, which progressively and adaptively learns over multiple stages to remove random outliers embedded in the initial consistency. Specifically, we propose an Adaptive Dynamic Graph Construction (ADGC) module to construct a global topological graph with a high inlier ratio by calculating the affinity between the corresponding neighbors. Additionally, we also design a Feature Mapping Processing (FMP) block and Multi-Stage Prediction (MSP) block to improve the accuracy of subsequent local neighborhood information aggregation, as well as information loss due to channel dimension compression in the prediction phase. Experimental results in camera pose estimation, remote sensing image registration, and homography estimation demonstrate that the proposed PMA-Net performs better than other state-of-the-art methods on various public datasets. Code: https://github.com/XiaojieLi11/PMA-Net

Deep knowledge distillation: A self-mutual learning framework for traffic prediction

Traffic flow prediction in spatio-temporal networks is a crucial aspect of Intelligent Transportation Systems (ITS). Existing traffic flow forecasting methods, particularly those utilizing graph neural networks, encounter limitations. When processing large-scale graph data, the depth of these models can restrict their ability to effectively capture complex relationships and patterns. Additionally, these methods often focus mainly on local neighborhood information, which can limit their capability to recognize and analyze global relationships and patterns within the graph data. Therefore, we proposed a deep knowledge distillation model, tailored to effectively capture spatio-temporal patterns in traffic flow prediction. This model incorporates a bidirectional random walk process on a directed graph, enabling it to effectively capture both spatial and temporal dependencies. Utilizing a blend of mutual learning and self-distillation, our approach enhances the detection of spatio-temporal relationships within traffic data and improves the feature perception ability at both local and global levels. We tested our model on two real-world datasets, achieving notable improvements in prediction accuracy, especially for predictions within a one-hour timeframe. In comparison to the baseline model, our proposed model achieved accuracy improvements of 0.19 and 0.18 on the respective datasets. These results highlight the success of using mutual learning and self-distillation to transfer knowledge effectively within and between models and to improve the model’s capability in identifying and extracting features.

SEC: More Accurate Clustering Algorithm via Structural Entropy

As one of the most popular machine learning tools in the field of unsupervised learning, clustering has been widely used in various practical applications. While numerous methods have been proposed for clustering, a commonly encountered issue is that the existing clustering methods rely heavily on local neighborhood information during the optimization process, which leads to suboptimal performance on real-world datasets. Besides, most existing clustering methods use Euclidean distances or densities to measure the similarity between data points. This could constrain the effectiveness of the algorithms for handling datasets with irregular patterns. Thus, a key challenge is how to effectively capture the global structural information in clustering instances to improve the clustering quality. In this paper, we propose a new clustering algorithm, called SEC. This algorithm uses the global structural information extracted from an encoding tree to guide the clustering optimization process. Based on the relation between data points in the instance, a sparse graph of the clustering instance can be constructed. By leveraging the sparse graph constructed, we propose an iterative encoding tree method, where hierarchical abstractions of the encoding tree are iteratively extracted as new clustering features to obtain better clustering results. To avoid the influence of easily misclustered data points located on the boundaries of the clustering partitions, which we call "fringe points", we propose an iterative pre-deletion and reassignment technique such that the algorithm can delete and reassign the "fringe points" to obtain more resilient and precise clustering results. Empirical experiments on both synthetic and real-world datasets demonstrate that our proposed algorithm outperforms state-of-the-art clustering methods and achieves better clustering performances. On average, the clustering accuracy (ACC) is increased by 1.7% and the normalized mutual information (NMI) by 7.9% compared with the current state-of-the-art (SOTA) algorithm on synthetic datasets. On real-world datasets, our method outperforms other clustering methods with an average increase of 12.3% in ACC and 5.2% in NMI, respectively.

GraphSAGE++: Weighted Multi-scale GNN for Graph Representation Learning

Graph neural networks (GNNs) have emerged as a powerful tool in graph representation learning. However, they are increasingly challenged by over-smoothing as network depth grows, compromising their ability to capture and represent complex graph structures. Additionally, some popular GNN variants only consider local neighbor information during node updating, ignoring the global structural information and leading to inadequate learning and differentiation of graph structures. To address these challenges, we introduce a novel graph neural network framework, GraphSAGE++. Our model extracts the representation of the target node at each layer and then concatenates all layer weighted representations to obtain the final result. In addition, the strategies combining double aggregations with weighted concatenation are proposed, which significantly enhance the model’s discernment and preservation of structural information. Empirical results on various datasets demonstrate that GraphSAGE++ excels in vertex classification, link prediction, and visualization tasks, surpassing existing methods in effectiveness.

EEI-NET: EDGE-ENHANCED INTERPOLATION NETWORK FOR SEMANTIC SEGMENTATION OF HISTORICAL BUILDING POINT CLOUDS

Abstract. In recent years, the conservation research of historical buildings and cultural relics has received a lot of attention from the state and the people, which not only provides a deeper understanding of their historical value and cultural significance, but also promotes the expansion of conservation research to the three-dimensional level. In this context, the semantic segmentation of historical building components is particularly important, which can provide basic support for various historical building applications, such as research and study of historical buildings, repair and protection, and 3D fine reconstruction, etc. However, most of the current methods for semantic segmentation of point clouds of historical buildings suffer from the problems of not being able to fully exploit the local neighborhood information of point clouds and poor edge segmentation. Therefore, we propose a new deep learning semantic segmentation-based approach, which we call EEI-Net. It is an end-to-end deep neural network in which we designed an edge enhancement interpolation (EEI) module and an edge interaction classifier (EIC). The edge enhancement interpolation module performs edge enhancement interpolation by fusing multi-layer features between the encoder and decoder. The edge interaction classifier enables the interaction of edge information through information transfer between individual nodes. EEI-Net incorporates contextual features and better preserves and enhances the edge information of the point cloud. We conduct experiments on the constructed historical architecture dataset, and the results show that the proposed EEI-Net has better performance.

Point cloud semantic segmentation based on local feature fusion and multilayer attention network

AbstractSemantic segmentation from a three‐dimensional point cloud is vital in autonomous driving, computer vision, and augmented reality. However, current semantic segmentation does not effectively use the point cloud's local geometric features and contextual information, essential for improving segmentation accuracy. A semantic segmentation network that uses local feature fusion and a multilayer attention mechanism is proposed to address these challenges. Specifically, the authors designed a local feature fusion module to encode the geometric and feature information separately, which fully leverages the point cloud's feature perception and geometric structure representation. Furthermore, the authors designed a multilayer attention pooling module consisting of local attention pooling and cascade attention pooling to extract contextual information. Local attention pooling is used to learn local neighbourhood information, and cascade attention pooling captures contextual information from deeper local neighbourhoods. Finally, an enhanced feature representation of important information is obtained by aggregating the features from the two deep attention pooling methods. Extensive experiments on large‐scale point‐cloud datasets Stanford 3D large‐scale indoor spaces and SemanticKITTI indicate that authors network shows excellent advantages over existing representative methods regarding local geometric feature description and global contextual relationships.

Kernel possibilistic fuzzy c-means clustering algorithm based on morphological reconstruction and membership filtering

A lot of research has been conducted on many variants of the Fuzzy c-means (FCM) clustering algorithm incorporating local spatial neighborhood information to improve segmentation accuracy and robustness to noise. Among these variants, a fast and robust FCM (FRFCM) clustering algorithm performs fast and robustly to noise for both grayscale and color images. Though, FRFCM is fast but segmentation performance needs improvement. This work presents an improved variant of the FRFCM algorithm, based on the kernel metric and possibilistic fuzzy c-means approach. The proposed method named Kernel Possibilistic Fast-Robust Fuzzy c-means (KPFRFCM) algorithm overcomes the disadvantages of FRFCM i.e. the poor segmentation performance and less robustness to noise, for both grayscale and color images. Experiments performed on various types of images without noise and images degraded by different types of noises with different degrees, prove that proposed KPFRFCM is more efficient and more robust to noise when compared with existing state-of-the-art algorithms for image segmentation.

Two-step multi-view data classification based on dynamic Graph-ELM

This paper focuses on the classification problem of multi-view data, aiming to improve the classification accuracy of current algorithms on multi-view data. Previous multi-view classification algorithms are usually based on exploiting the complementarity of different views and fusing features from different views. A representative category is the graph-based method, which builds a graph matrix for each view, and then fuses the graph matrices of different views to obtain a unified graph. These methods have the following problems: firstly, the graph matrix is simply based on sample similarity usually; secondly, the learned graph matrix does not change dynamically; thirdly, the weight of the graph representation matrix for a single view cannot be learned in the unified graph matrix. Therefore, this paper designs a Two-step classification algorithm based on Dynamic Graph-ELM, called TSDGELM. In the TSDGELM, the dynamic Graph-ELM is used to obtain the graph representation matrix of each view to save the local neighbor information of the data, and then a joint graph learning algorithm is designed based on the GBS (Graph-Based System) mechanism to fuse the graph matrix of the single-view, and finally the united graph is input into the classifier. To evaluate the effectiveness of the proposed method in this work, we conduct a series of experiments on eight datasets, and the results demonstrate the superiority of the proposed method.

Matching of crushed highly decomposed granite particles using 3D SHOT descriptors

Owing to the complex morphology and the fragility of highly decomposed granite (HDG), it is still a challenge to track the breakage process of HDG particles within a deformed sample in geomechanics research. An HDG particle matching method is proposed in this paper for the first time to track and match HDG fragments with their mother particles from which they break off. First, the X-ray micro-computed tomography (μCT) images of HDG particles were acquired from a one-dimensional compression test and a certain number of HDG particles and fragments were selected and reconstructed. Second, the original and fracture surfaces of the fragments were identified and segmented. Then, the essential morphological features of the particles and fragments were represented with the key points. The neighbourhood information around the key points was encoded by the three-dimensional (3D) feature descriptor. Next, the mismatches were filtered out by the random sample consensus (RANSAC) algorithm based on the similarity of the local neighbourhood information. Finally, the influences of several parameters on the matching results were investigated. The capability and efficiency of the HDG particle matching method were demonstrated using the μCT data of ten HDG particles and 32 fragments that resulted from the crushing of these particles.

Application of local and global interval embedding algorithms in uncertain system processes

The data collected during the fault monitoring of industrial processes are often uncertain because of measurement noise, sensor error, and complex working environments. Currently, most of this uncertainty is transformed into interval-value data for processing. However, the traditional interval fault monitoring method (TIFMM) only considers the global information of the interval-valued data and only rarely considers the local neighborhood information, which can characterize the topological relationship between the interval-valued data points, so that the sample points with some nonlinear relationship still maintain this nonlinear relationship after dimensionality reduction. Therefore, to take into account both local and global information, a new interval-valued data projection algorithm called local and global interval embedding algorithms (LGIEA) is proposed in this paper. First, the measurement error estimation method, based on the principle of reasonable granularity, converts the imprecise variable measurement values into interval-value data that can reflect uncertainty. Then, an interval-matrix feature extraction method based on local and global information fusion, proposed in this paper, extracts process features by minimizing local scattering and maximizing global scattering of interval-valued data. Thereby, it preserves the optimal local neighborhood and global information in low-dimensional space. In addition, four statistical indicators and corresponding variable-contribution graphs have been defined to use online fault monitoring and fault variable identification. Finally, the superiority of the LGIEA was verified using simulation data of the Tennessee Eastman process and actual operating data of a shearer in an inclined-ditch coal mine. The experimental results show that, Compared with TIFMM such as midpoint-radius principal component analysis (PCA), complete information PCA and spectral radius-based interval PCA, proposed method significantly reduces the error alarm rate and the false monitoring rate and improves the accuracy of data classification. It also has better fault identification accuracy and robustness.

Analysis of Adaptive Morphology for Visualization of Information Communication in Advertising Design

Abstract This paper uses adaptive morphology to construct an adaptive elliptic structure element that changes adaptively according to the image’s local neighborhood information. To calculate convolution, a Gaussian kernel with the outer product of the image gradient is used to achieve the diffusion of the linear matrix values of the initial matrix. The spatial distance of the pixels determines the size of the structural elements, and the pixel points that do not exceed the threshold are selected. The adaptive morphology method resulted in a 27.7 difficulty level for visual communication design, and the algorithm’s accuracy was 87.18%. Advertising design can break through dimensional limitations in visual space through adaptive morphology to generate interaction, indicating a path for development.

Locality Robust Domain Adaptation for cross-scene hyperspectral image classification

Domain adaptation (DA) has become a widely used technique for cross-scene hyperspectral image (HSI) classification. Most DA methods aim to learn a domain invariant subspace that reduce the domain discrepancy between source and target domains. However, some of them fail to explore the local manifold structure between different domains, while also neglecting the negative influences of abnormal features. To solve these problems, we propose a Locality Robust Domain Adaptation (LRDA) method for cross-domain data recognition. In LRDA, the statistical alignment is applied to reduce the domain-shift between the source and target domains. Then, LRDA combines the row-sparsity constraint and the discriminant regularization term to learn a robust projection matrix, while maintaining the discriminative capability of the matrix. Furthermore, a manifold regularization term is proposed to automatically learn the nearest neighbors and weights between two domains. The proposed LRDA not only reduces the discrepancy between two domains but also explores the local neighbor information between them. Experiment results on three HSI datasets illustrate that the proposed LRDA has better performance than other related methods.

Sensor data reduction with novel local neighborhood information granularity and rough set approach

Data description and data reduction are important issues in sensors data acquisition and rough sets based models can be applied in sensors data acquisition. Data description by rough set theory relies on information granularity, approximation methods and attribute reduction. The distribution of actual data is complex and changeable. The current model lacks the ability to distinguish different data areas leading to decision-making errors. Based on the above, this paper proposes a neighborhood decision rough set based on justifiable granularity. Firstly, the rough affiliation of the data points in different cases is given separately according to the samples in the neighborhood. Secondly, the original labels are rectified using pseudo-labels obtained from the label noise data that has been found. The new judgment criteria are proposed based on justifiable granularity, and the optimal neighborhood radius is optimized by the particle swarm algorithm. Finally, attribute reduction is performed on the basis of risky decision cost. Complex data can be effectively handled by the method, as evidenced by the experimental results.

Intuitionistic Fuzzy Weighted Least Squares Twin SVMs.

Fuzzy membership is an effective approach used in twin support vector machines (SVMs) to reduce the effect of noise and outliers in classification problems. Fuzzy twin SVMs (TWSVMs) assign membership weights to reduce the effect of outliers, however, it ignores the positioning of the input data samples and hence fails to distinguish between support vectors and noise. To overcome this issue, intuitionistic fuzzy TWSVM combined the concept of intuitionistic fuzzy number with TWSVMs to reduce the effect of outliers and distinguish support vectors from noise. Despite these benefits, TWSVMs and intuitionistic fuzzy TWSVMs still suffer from some drawbacks as: 1) the local neighborhood information is ignored among the data points and 2) they solve quadratic programming problems (QPPs), which is computationally inefficient. To overcome these issues, we propose a novel intuitionistic fuzzy weighted least squares TWSVMs for classification problems. The proposed approach uses local neighborhood information among the data points and also uses both membership and nonmembership weights to reduce the effect of noise and outliers. The proposed approach solves a system of linear equations instead of solving the QPPs which makes the model more efficient. We evaluated the proposed intuitionistic fuzzy weighted least squares TWSVMs on several benchmark datasets to show the efficiency of the proposed model. Statistical analysis is done to quantify the results statistically. As an application, we used the proposed model for the diagnosis of Schizophrenia disease.

Exploring attention mechanism for graph similarity learning

Graph similarity estimation is a challenging task due to the complex graph structure. Though important and well-studied, three key aspects are yet to be fully handled in a unified framework: (i) how to exploit the node embedding by leveraging both local spatial neighborhood information and the global context, (ii) how to effectively learn richer cross graph interactions from a pairwise node perspective and (iii) how to map the similarity matrix into a similarity score by exploiting the inherent structure in the similarity matrix. To solve these issues, we explore multiple attention mechanisms for graph similarity learning in this work. More specifically, we propose a unified graph similarity learning framework involving (i) a hybrid of graph convolution and graph self-attention for node embedding learning, (ii) a cross graph co-attention (GCA) module for graph interaction modeling, (iii) similarity-wise self-attention (SSA) module for graph similarity matrix alignment and (iv) graph similarity matrix learning for predicting the similarity scores. Extensive experimental results on three challenging benchmarks including LINUX, AIDS, and IMDBMulti demonstrate that the proposed NA-GSL performs favorably against state-of-the-art graph similarity estimation methods. The code is available at https://github.com/AlbertTan404/NA-GSL.

Clustering Heterogeneous Conformational Ensembles of Intrinsically Disordered Proteins with t-Distributed Stochastic Neighbor Embedding.

Intrinsically disordered proteins (IDPs) populate a range of conformations that are best described by a heterogeneous ensemble. Grouping an IDP ensemble into "structurally similar" clusters for visualization, interpretation, and analysis purposes is a much-desired but formidable task, as the conformational space of IDPs is inherently high-dimensional and reduction techniques often result in ambiguous classifications. Here, we employ the t-distributed stochastic neighbor embedding (t-SNE) technique to generate homogeneous clusters of IDP conformations from the full heterogeneous ensemble. We illustrate the utility of t-SNE by clustering conformations of two disordered proteins, Aβ42, and α-synuclein, in their APO states and when bound to small molecule ligands. Our results shed light on ordered substates within disordered ensembles and provide structural and mechanistic insights into binding modes that confer specificity and affinity in IDP ligand binding. t-SNE projections preserve the local neighborhood information, provide interpretable visualizations of the conformational heterogeneity within each ensemble, and enable the quantification of cluster populations and their relative shifts upon ligand binding. Our approach provides a new framework for detailed investigations of the thermodynamics and kinetics of IDP ligand binding and will aid rational drug design for IDPs.

MSaD-Net: A Mix Self-Attention Networks for 3D Point Cloud Denoising

In the process of acquiring 3D point cloud data, due to environmental interference or unstable scanning equipment, the acquired data often have noisy points. Recently, with the development of neural networks for point clouds, great progress has been made in deep learning-based point cloud denoising. However, most of the existing methods adopt a pointnet-like structure to predict point offsets. Simple pooling operation loses much important information, such as local neighborhood information, and global information. The loss of information makes the algorithm ineffective when dealing with some complex cases. In order to solve the above problems, we propose a self-attention-based point cloud denoising network architecture, through the Transformer structure, to establish long-range dependencies of the points. In addition, we propose a local information embedding module to fast select meaningful points and serve as the input of the Transformer. We also consider the correlation between channels of point cloud features and further introduce a channel attention module. Extensive experiments prove that our method outperforms existing methods, and maintains a high running speed.

Multiresolution discrete optimization registration method of ultrasound and magnetic resonance images based on key points

The registration of preoperative magnetic resonance (MR) images and intraoperative ultrasound (US) images is very important in the planning of brain tumor surgery and during surgery. Considering that the two-modality images have different intensity range and resolution, and the US images are degraded by lots of speckle noises, a self-similarity context (SSC) descriptor based on local neighborhood information was adopted to define the similarity measure. The ultrasound images were considered as the reference, the corners were extracted as the key points using three-dimensional differential operators, and the dense displacement sampling discrete optimization algorithm was adopted for registration. The whole registration process was divided into two stages including the affine registration and the elastic registration. In the affine registration stage, the image was decomposed using multi-resolution scheme, and in the elastic registration stage, the displacement vectors of key points were regularized using the minimum convolution and mean field reasoning strategies. The registration experiment was performed on the preoperative MR images and intraoperative US images of 22 patients. The overall error after affine registration was (1.57 ± 0.30) mm, and the average computation time of each pair of images was only 1.36 s; while the overall error after elastic registration was further reduced to (1.40 ± 0.28) mm, and the average registration time was 1.53 s. The experimental results show that the proposed method has prominent registration accuracy and high computational efficiency.