Pairwise Relationships Research Articles

DHGCN: Dynamic Hop Graph Convolution Network for Self-Supervised Point Cloud Learning

Recent works attempt to extend Graph Convolution Networks (GCNs) to point clouds for classification and segmentation tasks. These works tend to sample and group points to create smaller point sets locally and mainly focus on extracting local features through GCNs, while ignoring the relationship between point sets. In this paper, we propose the Dynamic Hop Graph Convolution Network (DHGCN) for explicitly learning the contextual relationships between the voxelized point parts, which are treated as graph nodes. Motivated by the intuition that the contextual information between point parts lies in the pairwise adjacent relationship, which can be depicted by the hop distance of the graph quantitatively, we devise a novel self-supervised part-level hop distance reconstruction task and design a novel loss function accordingly to facilitate training. In addition, we propose the Hop Graph Attention (HGA), which takes the learned hop distance as input for producing attention weights to allow edge features to contribute distinctively in aggregation. Eventually, the proposed DHGCN is a plug-and-play module that is compatible with point-based backbone networks. Comprehensive experiments on different backbones and tasks demonstrate that our self-supervised method achieves state-of-the-art performance. Our source codes are available at: https://github.com/Jinec98/DHGCN.

Abstract 1253: Pan-cancer comparative and integrative analyses of driver alterations using Japanese and international genomic databases

Abstract Recent large-scale pan-cancer genome projects have broadened our understanding of the cancer genome. However, these projects mainly consisted of White patients, with a limited focus on Asian populations. In addition, while pairwise relationships between driver alterations can provide valuable functional insights, the majority of these analyses are limited to those using The Cancer Genome Atlas (TCGA) dataset. Here we present a pan-cancer landscape of driver alterations in Japanese patients with advanced solid tumors using targeted sequencing data of 48,627 samples from the Center for Cancer Genomics and Advanced Therapeutics (C-CAT) in Japan. This data set had a greater number of samples from Asian-prevalent cancer types, such as intrahepatic cholangiocarcinoma and stomach adenocarcinoma, compared to the Genomics Evidence Neoplasia Information Exchange (GENIE) cohort. In the C-CAT cohort, the most frequently mutated gene was TP53 (56%), followed by KRAS (25%), APC (17%), and PIK3CA (12%). Gene fusions were detected in 1,166 samples (2%), including ERG fusions in prostate adenocarcinoma (n = 129), and RET and ALK fusions in lung adenocarcinoma (n = 30 and 32). At least one clinically actionable genetic lesion was found in 16% of patients, with well-differentiated thyroid cancer having the highest proportion. A comparative analysis of somatic mutation frequencies was conducted between the C-CAT cohort and the White population in the GENIE cohort. Among 352 driver gene-cancer type combinations, 14 and 4 were significantly higher and lower in the C-CAT cohort (> 10% difference with q-value < 0.01). Notably, TP53 mutation frequencies were elevated in the C-CAT cohort across 10 cancer types, suggesting racial differences in TP53 mutation frequencies. We then integrated C-CAT, GENIE, and TCGA data consisting of more than 150,000 patients to conduct a meta-analysis of co-occurring and mutually exclusive relationships between 1,790 cancer-driver combinations. This analysis not only validated 215 previously reported relationships but identified 484 novel ones. Interestingly, we found significantly co-occurring mutations within the epigenetic pathway, with 13 epigenetic regulators co-occurring in ten cancer types. Gene set enrichment analysis using RNA-sequencing data in TCGA revealed that accumulation of mutations in epigenetic regulators causes increased proliferation-related transcriptomic signatures. In addition, loss-of-function mutations in epigenetic drivers suppresses proliferation in wild-type cell lines, but this effect is reduced in cells with mutations in both the same and other epigenetic drivers. Our multi-cohort analyses uncover differences in the driver landscape between Asian and White populations, provides valuable resources for precision cancer medicine, and offers insights into epigenetic regulator-associated oncogenesis. Citation Format: Sara Horie, Yuki Saito, Yasunori Kogure, Kota Mizuno, Yuta Ito, Mariko Tabata, Koichi Murakami, Junji Koya, Keisuke Kataoka. Pan-cancer comparative and integrative analyses of driver alterations using Japanese and international genomic databases [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1253.

Attaining robust performance targets in data envelopment analysis with application to efficiency evaluation of airports

We consider a ratio-based efficiency analysis in view of achieving performance targets by Decision Making Units. The targets refer to the efficiency scores or ranks and the truth of pairwise preference relations. These outcomes can be instantiated for at least one or all feasible weights associated with inputs and outputs. We discuss Multiple Objective Optimization problems that minimize the required reductions of consumed resources or increases of produced results needed to reach the pre-defined aims. The proposed models offer great flexibility to the Decision Maker, who can indicate which factors should be modified and to which extent. The evolutionary optimization algorithm computes a set of solutions that exhibit the trade-offs between expected modifications of various factors. They can serve as the basis for interactively selecting the most preferred solution to be implemented in practice. The use of the proposed framework is illustrated in the problem of analyzing the efficiency of Polish airports.

External Debt and Economic Growth Relationship in Nigeria: A Reconsideration

External Debt and Economic Growth Relationship in Nigeria: A Reconsideration

Comparison of cerebrospinal fluid, plasma and neuroimaging biomarker utility in Alzheimer's disease.

Alzheimer's disease biomarkers are crucial to understanding disease pathophysiology, aiding accurate diagnosis and identifying target treatments. Although the number of biomarkers continues to grow, the relative utility and uniqueness of each is poorly understood as prior work has typically calculated serial pairwise relationships on only a handful of markers at a time. The present study assessed the cross-sectional relationships among 27 Alzheimer's disease biomarkers simultaneously and determined their ability to predict meaningful clinical outcomes using machine learning. Data were obtained from 527 community-dwelling volunteers enrolled in studies at the Charles F. and Joanne Knight Alzheimer Disease Research Center at Washington University in St Louis. We used hierarchical clustering to group 27 imaging, CSF and plasma measures of amyloid beta, tau [phosphorylated tau (p-tau), total tau t-tau)], neuronal injury and inflammation drawn from MRI, PET, mass-spectrometry assays and immunoassays. Neuropsychological and genetic measures were also included. Random forest-based feature selection identified the strongest predictors of amyloid PET positivity across the entire cohort. Models also predicted cognitive impairment across the entire cohort and in amyloid PET-positive individuals. Four clusters emerged reflecting: core Alzheimer's disease pathology (amyloid and tau), neurodegeneration, AT8 antibody-associated phosphorylated tau sites and neuronal dysfunction. In the entire cohort, CSF p-tau181/Aβ40lumi and Aβ42/Aβ40lumi and mass spectrometry measurements for CSF pT217/T217, pT111/T111, pT231/T231 were the strongest predictors of amyloid PET status. Given their ability to denote individuals on an Alzheimer's disease pathological trajectory, these same markers (CSF pT217/T217, pT111/T111, p-tau/Aβ40lumi and t-tau/Aβ40lumi) were largely the best predictors of worse cognition in the entire cohort. When restricting analyses to amyloid-positive individuals, the strongest predictors of impaired cognition were tau PET, CSF t-tau/Aβ40lumi, p-tau181/Aβ40lumi, CSF pT217/217 and pT205/T205. Non-specific CSF measures of neuronal dysfunction and inflammation were poor predictors of amyloid PET and cognitive status. The current work utilized machine learning to understand the interrelationship structure and utility of a large number of biomarkers. The results demonstrate that, although the number of biomarkers has rapidly expanded, many are interrelated and few strongly predict clinical outcomes. Examining the entire corpus of available biomarkers simultaneously provides a meaningful framework to understand Alzheimer's disease pathobiological change as well as insight into which biomarkers may be most useful in Alzheimer's disease clinical practice and trials.

Measuring the significance of higher-order dependency in networks

Higher-order networks (HONs), which go beyond the limitations of pairwise relation modeling by graphs, capture higher-order dependencies involving three or more components for various systems. As the number of potential higher-order dependencies increases exponentially with both network size and the order of dependency, it is of particular importance for HON models to balance their representation power against model complexity. In this study, we propose a method, significant k-order dependencies mining (SkDM), based on hypothesis testing and the Markov chain Monte Carlo (MCMC), to identify significant higher-order dependencies in real systems. Through synthetic clickstreams with elaborately designed higher-order dependencies, SkDM shows a powerful ability to correctly identify all significant dependencies at preset significance levels of α={0}{.01, 0}{.05, 0}{.10} , performing as the only method, in comparison to the state of the arts, that can robustly maintain the Type I error rate, and without generating any Type II error across all the experimental settings. We further apply the SkDM method to various empirical networks, including journal citations, air traffic, and email communications. Empirical results show that among those tested networks, only 6.03%, 1.47%, and 1.28% of all potential dependencies are of statistical significance ( α={0}{.01} ). The proposed SkDM method, therefore, provides an efficient tool for higher-order network analysis tasks at reduced computational complexity.

A topological deep learning framework for neural spike decoding

The brain’s spatial orientation system uses different neuron ensembles to aid in environment-based navigation. Two of the ways brains encode spatial information are through head direction cells and grid cells. Brains use head direction cells to determine orientation, whereas grid cells consist of layers of decked neurons that overlay to provide environment-based navigation. These neurons fire in ensembles where several neurons fire at once to activate a single head direction or grid. We want to capture this firing structure and use it to decode head direction and animal location from head direction and grid cell activity. Understanding, representing, and decoding these neural structures require models that encompass higher-order connectivity, more than the one-dimensional connectivity that traditional graph-based models provide. To that end, in this work, we develop a topological deep learning framework for neural spike train decoding. Our framework combines unsupervised simplicial complex discovery with the power of deep learning via a new architecture we develop herein called a simplicial convolutional recurrent neural network. Simplicial complexes, topological spaces that use not only vertices and edges but also higher-dimensional objects, naturally generalize graphs and capture more than just pairwise relationships. Additionally, this approach does not require prior knowledge of the neural activity beyond spike counts, which removes the need for similarity measurements. The effectiveness and versatility of the simplicial convolutional neural network is demonstrated on head direction and trajectory prediction via head direction and grid cell datasets.

Higher-order neurodynamical equation for simplex prediction

It is demonstrated that higher-order patterns beyond pairwise relations can significantly enhance the learning capability of existing graph-based models, and simplex is one of the primary form for graphically representing higher-order patterns. Predicting unknown (disappeared) simplices in real-world complex networks can provide us with deeper insights, thereby assisting us in making better decisions. Nevertheless, previous efforts to predict simplices suffer from two issues: (i) they mainly focus on 2- or 3-simplices, and there are few models available for predicting simplices of arbitrary orders, and (ii) they lack the ability to analyze and learn the features of simplices from the perspective of dynamics. In this paper, we present a Higher-order Neurodynamical Equation for Simplex Prediction of arbitrary order (HNESP), which is a framework that combines neural networks and neurodynamics. Specifically, HNESP simulates the dynamical coupling process of nodes in simplicial complexes through different relations (i.e., strong pairwise relation, weak pairwise relation, and simplex) to learn node-level representations, while explaining the learning mechanism of neural networks from neurodynamics. To enrich the higher-order information contained in simplices, we exploit the entropy and normalized multivariate mutual information of different sub-structures of simplices to acquire simplex-level representations. Furthermore, simplex-level representations and multi-layer perceptron are used to quantify the existence probability of simplices. The effectiveness of HNESP is demonstrated by extensive simulations on seven higher-order benchmarks. Experimental results show that HNESP improves the AUC values of the state-of-the-art baselines by an average of 8.32%. Our implementations will be publicly available at: https://github.com/jianruichen/HNESP.

How much freedom does an effectiveness metric really have?

AbstractIt is tempting to assume that because effectiveness metrics have free choice to assign scores to search engine result pages (SERPs) there must thus be a similar degree of freedom as to the relative order that SERP pairs can be put into. In fact that second freedom is, to a considerable degree, illusory. That is because if one SERP in a pair has been given a certain score by a metric, fundamental ordering constraints in many cases then dictate that the score for the second SERP must be either not less than, or not greater than, the score assigned to the first SERP. We refer to these fixed relationships as innate pairwise SERP orderings. Our first goal in this work is to describe and defend those pairwise SERP relationship constraints, and tabulate their relative occurrence via both exhaustive and empirical experimentation. We then consider how to employ such innate pairwise relationships in IR experiments, leading to a proposal for a new measurement paradigm. Specifically, we argue that tables of results in which many different metrics are listed for champion versus challenger system comparisons should be avoided; and that instead a single metric be argued for in principled terms, with any relationships identified by that metric then reinforced via an assessment of the innate relationship as to whether other metrics are likely to yield the same system‐versus‐system outcome.

Representation of the hierarchical and functional structure of an ambulatory network of medical consultations through Social Network Analysis, with an emphasis on the role of medical specialties.

Ambulatory Health Care Networks (Amb-HCN) are circuits of patient referral and counter-referral that emerge, explicitly or spontaneously, between doctors who provide care in their offices. Finding a meaningful analytical representation for the organic and hierarchical functioning of an Amb-HCN may have managerial and health policymaking implications. We aimed to characterize the structural and functional topology of an Amb-HCN of a private health insurance provider (PHIP) using objective metrics from graph theory. This is a cross-sectional quantitative study with a secondary data analysis study design. A Social Network Analysis (SNA) was conducted using office visits performed between April 1, 2021 and May 15, 2022, retrieved from secondary administrative claim databases from a PHIP in Belo Horizonte, Southeastern Brazil. Included were beneficiaries of a healthcare plan not restricting the location or physician caring for the patient. A directional and weighted network was constructed, where doctors were the vertices and patient referrals between doctors, within 7-45 days, were the network edges. Vertex-level SNA measures were calculated and grouped into three theoretical constructs: patient follow-up (aimed at assessing the doctor's pattern of patient follow-up); relationship with authorities (which assessed whether the doctor is an authority or contributes to his or her colleague's authority status); and centrality (aimed at positioning the doctor relative to the network graph). To characterize physician profiles within each dimension based on SNA metrics results, a K-means cluster analysis was conducted. The resulting physician clusters were assigned labels that sought to be representative of the observed values of the vertex metrics within the clusters. Overall, 666,263 individuals performed 3,863,222 office visits with 4,554 physicians. A total of 577 physicians (12.7%) had very low consultation productivity and contributed very little to the network (i.e., about 1.1% of all referrals made or received), being excluded from subsequent doctor profiles analysis. Cluster analysis found 951 (23.9%) doctors to be central in the graph and 1,258 (31.6%) to be peripheral; 883 (22.2%) to be authorities and 266 (6.7%) as seeking authorities; 3,684 (92.6%) mostly shared patients with colleagues, with patient follow-up intensities ranging from weak to strong. Wide profile dispersion was observed among specialties and, more interestingly, within specialties. Non-primary-care medical specialties (e.g., cardiology, endocrinology etc.) were associated with central profile in the graph, while surgical specialties predominated in the periphery, along with pediatrics. Only pediatrics was associated with strong and prevalent (i.e., low patient sharing pattern) follow-up. Many doctors from internal medicine and family medicine had unexpectedly weak and shared patient follow-up profiles. Doctor profiles exhibited pairwise relationships with each other and with the number of chronic comorbidities of the patients they treated. For example, physicians identified as authorities were frequently central and treated patients with more comorbidities. Ten medical communities were identified with clear territorial and specialty segregation. Viewing the Amb-HCN as a social network provided a topological and functional representation with potentially meaningful and actionable emerging insights into the most influential actors and specialties, functional hierarchies, factors that lead to self-constituted medical communities, and dispersion from expected patterns within medical specialties.

Correlation analysis of diabetes based on Copula.

The ratio of Triglyceride (TG) to high-density lipoprotein cholesterol (HDL-C) is a crucial indicator for diabetes diagnosis. This study utilizes the Copula function to model and fit the non-linear correlation among fasting blood glucose (Glu), glycosylated hemoglobin (HbA1C), and TG/HDL-C in patients with diabetes. The Copula function chosen for this study includes the two-dimensional Archimedes and Elliptical distribution family, as well as the multidimensional Vine Copula function, for fitting the data. The evaluation of the fitting effect is performed using the mean absolute error (MAE) and mean square error (MSE). The results indicate that the Clayton Copula exhibits the highest effectiveness in fitting the pairwise relationship between Glu and TG/HDL-C, as well as HbA1C and TG/HDL-C, displaying the smallest fitting error. Additionally, the Vine Copula function produces a satisfactory fit for the relationship among all three indicators. Compared to linear analysis methods, the Copula function more accurately depicts the correlation among these three types of indicators. Moreover, our findings indicate a stronger correlation in the lower tail between Glu and HbA1C, as well as TG/HDL-C, suggesting that the Copula function provides greater accuracy and applicability in depicting the relationship among these indicators. As a result, it can offer a more precise auxiliary diagnosis and serve as a valuable reference in clinical judgment.

Inference of Gene Regulatory Networks Based on Multi-view Hierarchical Hypergraphs.

Since gene regulation is a complex process in which multiple genes act simultaneously, accurately inferring gene regulatory networks (GRNs) is a long-standing challenge in systems biology. Although graph neural networks can formally describe intricate gene expression mechanisms, current GRN inference methods based on graph learning regard only transcription factor (TF)-target gene interactions as pairwise relationships, and cannot model the many-to-many high-order regulatory patterns that prevail among genes. Moreover, these methods often rely on limited prior regulatory knowledge, ignoring the structural information of GRNs in gene expression profiles. Therefore, we propose a multi-view hierarchical hypergraphs GRN (MHHGRN) inference model. Specifically, multiple heterogeneous biological information is integrated to construct multi-view hierarchical hypergraphs of TFs and target genes, using hypergraph convolution networks to model higher order complex regulatory relationships. Meanwhile, the coupled information diffusion mechanism and the cross-domain messaging mechanism facilitate the information sharing between genes to optimise gene embedding representations. Finally, a unique channel attention mechanism is used to adaptively learn feature representations from multiple views for GRN inference. Experimental results show that MHHGRN achieves better results than the baseline methods on the E. coli and S. cerevisiae benchmark datasets of the DREAM5 challenge, and it has excellent cross-species generalization, achieving comparable or better performance on scRNA-seq datasets from five mouse and two human cell lines.

Learning beyond-pairwise interactions enables the bottom–up prediction of microbial community structure

Understanding the assembly of multispecies microbial communities represents a significant challenge in ecology and has wide applications in agriculture, wastewater treatment, and human healthcare domains. Traditionally, studies on the microbial community assembly focused on analyzing pairwise relationships among species; however, neglecting higher-order interactions, i.e., the change of pairwise relationships in the community context, may lead to substantial deviation from reality. Herein, we have proposed a simple framework that incorporates higher-order interactions into a bottom-up prediction of the microbial community assembly and examined its accuracy using a seven-member synthetic bacterial community on a host plant, duckweed. Although the synthetic community exhibited emergent properties that cannot be predicted from pairwise coculturing results, our results demonstrated that incorporating information from three-member combinations allows the acceptable prediction of the community structure and actual interaction forces within it. This reflects that the occurrence of higher-order effects follows consistent patterns, which can be predicted even from trio combinations, the smallest unit of higher-order interactions. These results highlight the possibility of predicting, explaining, and understanding the microbial community structure from the bottom-up by learning interspecies interactions from simple beyond-pairwise combinations.

From local to global gene co-expression estimation using single-cell RNA-seq data.

In genomics studies, the investigation of gene relationships often brings important biological insights. Currently, the large heterogeneous datasets impose new challenges for statisticians because gene relationships are often local. They change from one sample point to another, may only exist in a subset of the sample, and can be nonlinear or even nonmonotone. Most previous dependence measures do not specifically target local dependence relationships, and the ones that do are computationally costly. In this paper, we explore a state-of-the-art network estimation technique that characterizes gene relationships at the single cell level, under the name of cell-specific gene networks. We first show that averaging the cell-specific gene relationship over a population gives a novel univariate dependence measure, the averaged Local Density Gap (aLDG), that accumulates local dependence and can detect any nonlinear, nonmonotone relationship. Together with a consistent nonparametric estimator, we establish its robustness on both the population and empirical levels. Then, we show that averaging the cell-specific gene relationship over mini-batches determined by some external structure information (eg, spatial or temporal factor) better highlights meaningful local structure change points. We explore the application of aLDG and its minibatch variant in many scenarios, including pairwise gene relationship estimation, bifurcating point detection in cell trajectory, and spatial transcriptomics structure visualization. Both simulations and real data analysis show that aLDG outperforms existing ones.

On Unicyclic Graphs with Minimum Graovac–Ghorbani Index

In discrete mathematics, graph theory is the study of graphs, which are mathematical structures used to model pairwise relations between objects. Chemical graph theory is concerned with non-trivial applications of graph theory to the solution of molecular problems. Its main goal is to use numerical invariants to reduce the topological structure of a molecule to a single number that characterizes its properties. Topological indices are numerical invariants associated with the chemical constitution, for the purpose of the correlation of chemical structures with various physical properties, chemical reactivity, or biological activity. They have found important application in predicting the behavior of chemical substances. The Graovac–Ghorbani (ABCGG) index is a topological descriptor that has improved predictive potential compared to analogous descriptors. It is used to model both the boiling point and melting point of molecules and is applied in the pharmaceutical industry. In the recent years, the number of publications on its mathematical properties has increased. The aim of this work is to partially solve an open problem, namely to find the structure of unicyclic graphs that minimize the ABCGG index. We characterize unicyclic graphs with even girth that minimize the ABCGG index, while we also present partial results for odd girths. As an auxiliary result, we compare the ABCGG indices of paths and cycles with an odd number of vertices.

Clustering on hierarchical heterogeneous data with prior pairwise relationships

BackgroundClustering is a fundamental problem in statistics and has broad applications in various areas. Traditional clustering methods treat features equally and ignore the potential structure brought by the characteristic difference of features. Especially in cancer diagnosis and treatment, several types of biological features are collected and analyzed together. Treating these features equally fails to identify the heterogeneity of both data structure and cancer itself, which leads to incompleteness and inefficacy of current anti-cancer therapies.ObjectivesIn this paper, we propose a clustering framework based on hierarchical heterogeneous data with prior pairwise relationships. The proposed clustering method fully characterizes the difference of features and identifies potential hierarchical structure by rough and refined clusters.ResultsThe refined clustering further divides the clusters obtained by the rough clustering into different subtypes. Thus it provides a deeper insight of cancer that can not be detected by existing clustering methods. The proposed method is also flexible with prior information, additional pairwise relationships of samples can be incorporated to help to improve clustering performance. Finally, well-grounded statistical consistency properties of our proposed method are rigorously established, including the accurate estimation of parameters and determination of clustering structures.ConclusionsOur proposed method achieves better clustering performance than other methods in simulation studies, and the clustering accuracy increases with prior information incorporated. Meaningful biological findings are obtained in the analysis of lung adenocarcinoma with clinical imaging data and omics data, showing that hierarchical structure produced by rough and refined clustering is necessary and reasonable.

Stability and Generalization of Hypergraph Collaborative Networks

Graph neural networks have been shown to be very effective in utilizing pairwise relationships across samples. Recently, there have been several successful proposals to generalize graph neural networks to hypergraph neural networks to exploit more complex relationships. In particular, the hypergraph collaborative networks yield superior results compared to other hypergraph neural networks for various semi-supervised learning tasks. The collaborative network can provide high quality vertex embeddings and hyperedge embeddings together by formulating them as a joint optimization problem and by using their consistency in reconstructing the given hypergraph. In this paper, we aim to establish the algorithmic stability of the core layer of the collaborative network and provide generalization guarantees. The analysis sheds light on the design of hypergraph filters in collaborative networks, for instance, how the data and hypergraph filters should be scaled to achieve uniform stability of the learning process. Some experimental results on real-world datasets are presented to illustrate the theory.

Contrastive Graph Similarity Networks

Graph similarity learning is a significant and fundamental issue in the theory and analysis of graphs, which has been applied in a variety of fields, including object tracking, recommender systems, similarity search, and so on. Recent methods for graph similarity learning that utilize deep learning typically share two deficiencies: (1) they leverage graph neural networks as backbones for learning graph representations but have not well captured the complex information inside data, and (2) they employ a cross-graph attention mechanism for graph similarity learning, which is computationally expensive. Taking these limitations into consideration, a method for graph similarity learning is devised in this study, namely, Contrastive Graph Similarity Network (CGSim). To enhance graph similarity learning, CGSim makes use of the complementary information of two input graphs and captures pairwise relations in a contrastive learning framework. By developing a dual contrastive learning module with a node-graph matching and a graph-graph matching mechanism, our method significantly reduces the quadratic time complexity for cross-graph interaction modeling to linear time complexity. Jointly learning in an end-to-end framework, the graph representation embedding module and the well-designed contrastive learning module can be beneficial to one another. A comprehensive series of experiments indicate that CGSim outperforms state-of-the-art baselines on six datasets and significantly reduces the computational cost, which demonstrates our CGSim model’s superiority over other baselines.

Learning spatial–temporal pairwise and high-order relationships for short-term passenger flow prediction in urban rail transit

Short-term passenger flow prediction (STPFP) helps ease traffic congestion and optimize urban rail transit (URT) system resource allocation. Although graph-based models have been widely used in STPFP because of their ability to model complex spatial–temporal dependencies, accurate STPFP still faces challenges. The graph-based methods exploit pairwise stations to model spatial dependencies without considering the non-pairwise high-order relationships between multiple stations. Besides, the event’s occurrence will make the passenger flow at a particular moment have a more profound impact on the passenger flow at a future moment. To solve these issues, a spatial–temporal hypergraph attention recurrent network (STHGARN) method is proposed for STPFP. Specifically, STHGARN first models passenger flow data’s recent, daily, and weekly characteristics through three parallel components. Then, each parallel component uses the hypergraph attention recurrent network (HGARN) cell to jointly learn the pairwise and high-order relationships of multiple stations. Finally, a temporal Hawkes attention mechanism is constructed and concatenated after each parallel HGARN cell to simulate the influence of events and capture the historical key moments that affect the future passenger flow. Numerous experiments on three metro ridership datasets show that STHGARN is superior to the most advanced methods, and good prediction results can lay the foundation for the operation and management of URT.

Meta Pairwise Relationship Distillation for Unsupervised Person Re-Identification.

Unsupervised person re-identification (Re-ID) is challenging due to the lack of ground-truth labels. Most existing methods rely on pseudo labels estimated via iterative clustering and thus are highly susceptible to performance penalties incurred by the inaccurate estimated number of clusters. Alternatively, we utilize the sample pairs with pairwise pseudo labels to guide the feature learning to avoid the dilemma of determining cluster numbers. In this article, we propose a meta pairwise relationship distillation (MPRD) method that incorporates a graph convolutional network (GCN) to provide high-fidelity pairwise relationships to supervise the model training. A small amount of metadata with very-confidence pairwise relationships and the unlabeled pairs with the provided pseudo pairwise relationships participate in the GCN training. Besides, we introduce a hard sample deduction (HSD) module to timely mine the sample pairs with error-prone pairwise pseudo labels to mitigate the misled optimization by noisy labels. Furthermore, since the features of each positive pair represent the same person, we design a positive pair alignment (PPA) module to reduce the redundant information in each feature, which is achieved by minimizing the difference between each positive pair's feature distributions. Extensive experiments on the Market-1501, DukeMTMC-reID, and MSMT17 datasets show that our method outperforms the state-of-the-art unsupervised methods.