Multi-view Model Research Articles

A multi-view prognostic model for diffuse large B-cell lymphoma based on kernel canonical correlation analysis and support vector machine

Background and objectivePositron emission tomography/computed tomography (PET/CT) is recommended as the standard imaging modality for diffuse large B-cell lymphoma (DLBCL) staging. However, many studies have neglected the role of patients' prognostic factors with respect to imaging PET/CT of quantitative features. In this paper, a multi-view learning (MVL) model is established to make full use of both clinical and imaging data to predict the prognosis of DLBCL patients and thereby assist doctors in decision-making.MethodsFeature engineering, including feature extraction, feature screening by recursive feature elimination, and dimensionality reduction by principal component analysis, are successively performed on the clinical data and imaging data of the research subjects to obtain the study data. After dividing the data into training and test sets, an instance weighting method is applied to the training data. Subsequently, kernel mapping is performed on the imaging features and clinical features separately, and this kernel mapping is processed in the new kernel feature space using kernel canonical correlation analysis (KCCA). Lastly, model training is performed on the obtained common kernel subspace using a support vector machine (SVM). The final overall model, named SVM-2view-KCCA (SVM-2 K), was compared with three other multi-view models (Ensemble-SVM, Multi-view maximum entropy discrimination, and canonical correlation analysis). The performance of the model was evaluated on the test data with respect to several dichotomous metrics: accuracy, sensitivity, F1 score, the area under the curve (AUC), and G-mean.ResultsThe SVM model improved AUC by 10.5%, sensitivity by 11.9%, accuracy by 9.8%, F1 score by 9.2%, and G-mean by 7.8% for the DLBCL test data after feature engineering based on dimensionality reduction and instance weighting. In the performance comparison of single-view learning models, the SVM-based integration of clinical and imaging features achieved the best overall performance (AUC = 86.3%, accuracy = 91.6%, sensitivity = 83.2%, F1 = 85.7%, and G-mean = 86.1%). In the comparison of MVL models, SVM-2 K achieved the best overall performance (AUC = 92.1%, accuracy = 96.9%, sensitivity = 90.9%, F1 = 92.8%, and G-mean = 91.4%), and the performance of each MVL model was better than that of the best single-view learning model.ConclusionsMVL models outperformed single-view learning models. Of the MVL models, the proposed SVM-2 K achieved the best overall performance and could accurately predict patient prognosis.

EDDINet: Enhancing drug–drug interaction prediction via information flow and consensus constrained multi-graph contrastive learning

Predicting drug–drug interactions (DDIs) is crucial for understanding and preventing adverse drug reactions (ADRs). However, most existing methods inadequately explore the interactive information between drugs in a self-supervised manner, limiting our comprehension of drug–drug associations. This paper introduces EDDINet: Enhancing Drug-Drug Interaction Prediction via Information Flow and Consensus-Constrained Multi-Graph Contrastive Learning for precise DDI prediction. We first present a cross-modal information-flow mechanism to integrate diverse drug features, enriching the structural insights conveyed by the drug feature vector. Next, we employ contrastive learning to filter various biological networks, enhancing the model’s robustness. Additionally, we propose a consensus regularization framework that collaboratively trains multi-view models, producing high-quality drug representations. To unify drug representations derived from different biological information, we utilize an attention mechanism for DDI prediction. Extensive experiments demonstrate that EDDINet surpasses state-of-the-art unsupervised models and outperforms some supervised baseline models in DDI prediction tasks. Our approach shows significant advantages and holds promising potential for advancing DDI research and improving drug safety assessments. Our codes are available at: https://github.com/95LY/EDDINet_code.

Multi-view graph contrastive representation learning for bundle recommendation

Bundle recommendation can recommend a collection of associated items that can be consumed together to a user rather than recommending these items separately, making it extremely suitable for some scenarios such as product bundle recommendation and game bundle recommendation. Recent bundle recommendation approaches consider auxiliary data to mitigate sparse user-bundle interactions. However, these approaches obtain the node embeddings directly from the established user-bundle graph and do not explicitly exploit the relationships between users (bundles) when constructing recommendation models. Moreover, bundle recommendation approaches based on graph contrastive learning usually construct contrastive views by randomly discarding nodes (edges) in the graph, while discarding some essential nodes or edges will destroy the structure of the original graph, thereby deteriorating the quality of the learned node embeddings. Aiming at these limitations, we propose a bundle recommendation approach based on multi-view graph contrastive representation learning. First, we present a multi-view modeling method to model the relations between entities as several views from different perspectives. These views serve as inputs of graph neural networks for graph representation learning and provide contrastive views for the contrastive learning tasks. Second, we propose a novel framework for bundle recommendation. This framework obtains the user (bundle) embeddings from different views by performing multi-view graph representation learning and enhances the learned user and bundle embeddings through a two-level contrastive learning strategy. On this basis, the enhanced user (bundle) embeddings are fused for prediction. Finally, we design a joint optimization objective to optimize the model parameters, combining the prediction loss that supports multiple negative samples and the contrastive losses. Experiments on the Netease and Youshu datasets reveal that our approach outperforms the state-of-the-art (SOTA) baselines. Furthermore, the average improvements of Recall@K and NDCG@K of our approach over the SOTA baselines are approximately 3.38% and 2.80% on Netease and 3.94% and 4.84% on Youshu.

Discriminative Deep Canonical Correlation Analysis for Multi-View Data.

Over the past few years, multimodal data analysis has emerged as an inevitable method for identifying sample categories. In the multi-view data classification problem, it is expected that the joint representation should include the supervised information of sample categories so that the similarity in the latent space implies the similarity in the corresponding concepts. Since each view has different statistical properties, the joint representation should be able to encapsulate the underlying nonlinear data distribution of the given observations. Another important aspect is the coherent knowledge of the multiple views. It is required that the learning objective of the multi-view model efficiently captures the nonlinear correlated structures across different modalities. In this context, this article introduces a novel architecture, termed discriminative deep canonical correlation analysis (D2CCA), for classifying given observations into multiple categories. The learning objective of the proposed architecture includes the merits of generative models to identify the underlying probability distribution of the given observations. In order to improve the discriminative ability of the proposed architecture, the supervised information is incorporated into the learning objective of the proposed model. It also enables the architecture to serve as both a feature extractor as well as a classifier. The theory of CCA is integrated with the objective function so that the joint representation of the multi-view data is learned from maximally correlated subspaces. The proposed framework is consolidated with corresponding convergence analysis. The efficacy of the proposed architecture is studied on different domains of applications, namely, object recognition, document classification, multilingual categorization, face recognition, and cancer subtype identification with reference to several state-of-the-art methods.

ScMGATGRN: a multiview graph attention network-based method for inferring gene regulatory networks from single-cell transcriptomic data.

The gene regulatory network (GRN) plays a vital role in understanding the structure and dynamics of cellular systems, revealing complex regulatory relationships, and exploring disease mechanisms. Recently, deep learning (DL)-based methods have been proposed to infer GRNs from single-cell transcriptomic data and achieved impressive performance. However, these methods do not fully utilize graph topological information and high-order neighbor information from multiple receptive fields. To overcome those limitations, we propose a novel model based on multiview graph attention network, namely, scMGATGRN, to infer GRNs. scMGATGRN mainly consists of GAT, multiview, and view-level attention mechanism. GAT can extract essential features of the gene regulatory network. The multiview model can simultaneously utilize local feature information and high-order neighbor feature information of nodes in the gene regulatory network. The view-level attention mechanism dynamically adjusts the relative importance of node embedding representations and efficiently aggregates node embedding representations from two views. To verify the effectiveness of scMGATGRN, we compared its performance with 10 methods (five shallow learning algorithms and five state-of-the-art DL-based methods) on seven benchmark single-cell RNA sequencing (scRNA-seq) datasets from five cell lines (two in human and three in mouse) with four different kinds of ground-truth networks. The experimental results not only show that scMGATGRN outperforms competing methods but also demonstrate the potential of this model in inferring GRNs. The code and data of scMGATGRN are made freely available on GitHub (https://github.com/nathanyl/scMGATGRN).

Integrated multi-view modeling for reliable machine learning-intensive software engineering

Development of machine learning (ML) systems differs from traditional approaches. The probabilistic nature of ML leads to a more experimentative development approach, which often results in a disparity between the quality of ML models with other aspects such as business, safety, and the overall system architecture. Herein the Multi-view Modeling Framework for ML Systems (M3S) is proposed as a solution to this problem. M3S provides an analysis framework that integrates different views. It is supported by an integrated metamodel to ensure the connection and consistency between different models. To facilitate the experimentative nature of ML training, M3S provides an integrated platform between the modeling environment and the ML training pipeline. M3S is validated through a case study and a controlled experiment. M3S shows promise, but future research needs to confirm its generality.

Dual-stream multi-dependency graph neural network enables precise cancer survival analysis

Histopathology image-based survival prediction aims to provide a precise assessment of cancer prognosis and can inform personalized treatment decision-making in order to improve patient outcomes. However, existing methods cannot automatically model the complex correlations between numerous morphologically diverse patches in each whole slide image (WSI), thereby preventing them from achieving a more profound understanding and inference of the patient status. To address this, here we propose a novel deep learning framework, termed dual-stream multi-dependency graph neural network (DM-GNN), to enable precise cancer patient survival analysis. Specifically, DM-GNN is structured with the feature updating and global analysis branches to better model each WSI as two graphs based on morphological affinity and global co-activating dependencies. As these two dependencies depict each WSI from distinct but complementary perspectives, the two designed branches of DM-GNN can jointly achieve the multi-view modeling of complex correlations between the patches. Moreover, DM-GNN is also capable of boosting the utilization of dependency information during graph construction by introducing the affinity-guided attention recalibration module as the readout function. This novel module offers increased robustness against feature perturbation, thereby ensuring more reliable and stable predictions. Extensive benchmarking experiments on five TCGA datasets demonstrate that DM-GNN outperforms other state-of-the-art methods and offers interpretable prediction insights based on the morphological depiction of high-attention patches. Overall, DM-GNN represents a powerful and auxiliary tool for personalized cancer prognosis from histopathology images and has great potential to assist clinicians in making personalized treatment decisions and improving patient outcomes.

PathIntegrate: Multivariate modelling approaches for pathway-based multi-omics data integration.

As terabytes of multi-omics data are being generated, there is an ever-increasing need for methods facilitating the integration and interpretation of such data. Current multi-omics integration methods typically output lists, clusters, or subnetworks of molecules related to an outcome. Even with expert domain knowledge, discerning the biological processes involved is a time-consuming activity. Here we propose PathIntegrate, a method for integrating multi-omics datasets based on pathways, designed to exploit knowledge of biological systems and thus provide interpretable models for such studies. PathIntegrate employs single-sample pathway analysis to transform multi-omics datasets from the molecular to the pathway-level, and applies a predictive single-view or multi-view model to integrate the data. Model outputs include multi-omics pathways ranked by their contribution to the outcome prediction, the contribution of each omics layer, and the importance of each molecule in a pathway. Using semi-synthetic data we demonstrate the benefit of grouping molecules into pathways to detect signals in low signal-to-noise scenarios, as well as the ability of PathIntegrate to precisely identify important pathways at low effect sizes. Finally, using COPD and COVID-19 data we showcase how PathIntegrate enables convenient integration and interpretation of complex high-dimensional multi-omics datasets. PathIntegrate is available as an open-source Python package.

Multi-Domain Deep Learning from a Multi-View Perspective for Cross-Border E-commerce Search

Building click-through rate (CTR) and conversion rate (CVR) prediction models for cross-border e-commerce search requires modeling the correlations among multi-domains. Existing multi-domain methods would suffer severely from poor scalability and low efficiency when number of domains increases. To this end, we propose a Domain-Aware Multi-view mOdel (DAMO), which is domain-number-invariant, to effectively leverage cross-domain relations from a multi-view perspective. Specifically, instead of working in the original feature space defined by different domains, DAMO maps everything to a new low-rank multi-view space. To achieve this, DAMO firstly extracts multi-domain features in an explicit feature-interactive manner. These features are parsed to a multi-view extractor to obtain view-invariant and view-specific features. Then a multi-view predictor inputs these two sets of features and outputs view-based predictions. To enforce view-awareness in the predictor, we further propose a lightweight view-attention estimator to dynamically learn the optimal view-specific weights w.r.t. a view-guided loss. Extensive experiments on public and industrial datasets show that compared with state-of-the-art models, our DAMO achieves better performance with lower storage and computational costs. In addition, deploying DAMO to a large-scale cross-border e-commence platform leads to 1.21%, 1.76%, and 1.66% improvements over the existing CGC-based model in the online AB-testing experiment in terms of CTR, CVR, and Gross Merchandises Value, respectively.

Decoupled representation for multi-view learning

Learning multi-view data is a central topic for advanced deep model applications. Existing efforts mainly focus on exploring shared information to maximize the consensus among all the views. However, after reasonably discarding superfluous task-irrelevant noise, the view-specific information is equally essential to downstream tasks. In this paper, we propose to decouple the multi-view representation learning into the shared and specific information extractions with parallel branches, and seamlessly adopt feature fusion in end-to-end models. The common feature is obtained based on the view-agnostic contrastive learning and view-discriminative training to minimize the discrepancy within the views. Simultaneously, the specific feature is learned with orthogonality constraints to minimize the view-level correlation. Besides, the semantic information in the features is reserved with supervised training. After disentangling the representations, we fuse the mutually complementary common and specific features for downstream tasks. Particularly, we provide a theoretical explanation for our method from an information bottleneck perspective. Compared with state-of-the-art multi-view models on benchmark datasets, we empirically demonstrate the advantage of our method in several downstream tasks, such as ordinary classification and few-shot learning. In addition, extensive experiments validate the robustness and transferability of our approach, when applying the learned representation on the source dataset to several target datasets.

Self-avatar: Monocular 3D human reconstruction from RGB image

3D human position and shape estimate are crucial in many computer vision applications. Despite the fact that there are numerous deep learning techniques designed to handle this problem, they frequently only use training networks with RGB images from a single point of view. In this paper, a unique approach to solve this issue is proposed by combining a regression-based multi-view picture learning loop with an optimization-based multi-view model. This is because some public datasets are collected by multi-view camera systems. A parameterized human body model's position and shape parameters are initially deduced by a convolutional neural network (CNN) from multi-view photos. This work then introduces an enhanced multi-view optimization method called MV-SMPLify, which aligns the SMPL model with multi-view images by using the regressed pose and shape as beginning values. Following that, the CNN model's training can be monitored using the optimum parameters. The Self-avatar project as a whole is a self-supervised framework that combines the advantages of both the CNN method and the optimization-based strategy. Additionally, the use of multi-view photos improves thorough supervision during training. This methodology outperforms earlier methods in a variety of ways, according to qualitative and quantitative testing using open datasets.

ScHybridBERT: integrating gene regulation and cell graph for spatiotemporal dynamics in single-cell clustering.

Graph learning models have received increasing attention in the computational analysis of single-cell RNA sequencing (scRNA-seq) data. Compared with conventional deep neural networks, graph neural networks and language models have exhibited superior performance by extracting graph-structured data from raw gene count matrices. Established deep neural network-based clustering approaches generally focus on temporal expression patterns while ignoring inherent interactions at gene-level as well as cell-level, which could be regarded as spatial dynamics in single-cell data. Both gene-gene and cell-cell interactions are able to boost the performance of cell type detection, under the framework of multi-view modeling. In this study, spatiotemporal embedding and cell graphs are extracted to capture spatial dynamics at the molecular level. In order to enhance the accuracy of cell type detection, this study proposes the scHybridBERT architecture to conduct multi-view modeling of scRNA-seq data using extracted spatiotemporal patterns. In this scHybridBERT method, graph learning models are employed to deal with cell graphs and the Performer model employs spatiotemporal embeddings. Experimental outcomes about benchmark scRNA-seq datasets indicate that the proposed scHybridBERT method is able to enhance the accuracy of single-cell clustering tasks by integrating spatiotemporal embeddings and cell graphs.

AFSleepNet: Attention-based multi-view feature fusion framework for pediatric sleep staging.

The widespread prevalence of sleep problems in children highlights the importance of timely and accurate sleep staging in the diagnosis and treatment of pediatric sleep disorders. However, most existing sleep staging methods rely on one-dimensional raw polysomnograms or two-dimensional spectrograms, which omit critical details due to single-view processing. This shortcoming is particularly apparent in pediatric sleep staging, where the lack of a specialized network fails to meet the needs of precision medicine. Therefore, we introduce AFSleepNet, a novel attention-based multi-view feature fusion network tailored for pediatric sleep analysis. The model utilizes multimodal data (EEG, EOG, EMG), combining one-dimensional convolutional neural networks to extract time-invariant features and bidirectional-long-short-term memory to learn the transition rules among sleep stages, as well as employing short-time Fourier transform to generate two-dimensional spectral maps. This network employs a fusion method with self-attention mechanism and innovative pre-training strategy. This strategy can maintain the feature extraction capabilities of AFSleepNet from different views, enhancing the robustness of the multi-view model while effectively preventing model overfitting, thereby achieving efficient and accurate automatic sleep stage analysis. A "leave-one-subject-out" cross-validation on CHAT and clinical datasets demonstrated the excellent performance of AFSleepNet, with mean accuracies of 87.5% and 88.1%, respectively. Superiority over existing methods improves the accuracy and reliability of pediatric sleep staging.

Single-Frame-Based Deep View Synchronization for Unsynchronized Multicamera Surveillance.

Multicamera surveillance has been an active research topic for understanding and modeling scenes. Compared to a single camera, multicameras provide larger field-of-view and more object cues, and the related applications are multiview counting, multiview tracking, 3-D pose estimation or 3-D reconstruction, and so on. It is usually assumed that the cameras are all temporally synchronized when designing models for these multicamera-based tasks. However, this assumption is not always valid, especially for multicamera systems with network transmission delay and low frame rates due to limited network bandwidth, resulting in desynchronization of the captured frames across cameras. To handle the issue of unsynchronized multicameras, in this article, we propose a synchronization model that works in conjunction with existing deep neural network (DNN)-based multiview models, thus avoiding the redesign of the whole model. We consider two variants of the model, based on where in the pipeline the synchronization occurs, scene-level synchronization and camera-level synchronization. The view synchronization step and the task-specific view fusion and prediction step are unified in the same framework and trained in an end-to-end fashion. Our view synchronization models are applied to different DNNs-based multicamera vision tasks under the unsynchronized setting, including multiview counting and 3-D pose estimation, and achieve good performance compared to baselines.

Predicting Urban Region Heat via Learning Arrive-Stay-Leave Behaviors of Private Cars

Urban region heat refers to the extent of which people congregate in various regions when they travel to and stay in a specified place. Predicting urban region heat facilitates broad applications ranging from location-based services to intelligent transportation management. The region heat is essentially characterized by the ‘arrive-stay-leave (ASL)’ behaviors, while it is a challenging task to well capture the spatial-temporal evolution of region heat since the following issues remain: i) ASL behaviors of private cars is usually heterogeneous resulting in a hierarchical distribution of region heat. ii) Urban region heat contains complex spatial-temporal correlations hidden in ASL behaviors and how to collaboratively integrate them is challenging. To address these challenges, we propose a Hierarchical Spatial-Temporal Network (HierSTNet) to forecast urban region heat, which contains two representations, namely, grid region from micro perspective and node region from macro perspective. For the grids, three-dimension spatial and temporal convolutional network (3D-STCNN) is proposed to model multi-scale properties in temporal dimension of ASL behaviors. For the nodes, multi-head graph attention networks are utilized to model the periodicity and spatial heterogeneity among macro region. Hierarchical structures are designed for multi-view modeling spatial-temporal distribution of ASL behaviors, by which they capture small-scale features in micro regions and embeds the global representation into graph propagation. Finally, we design an interaction decoder layer to integrate the external factors and aggregate spatial-temporal information across hierarchical structures. Extensive experiments based on real-world private car trajectory dataset demonstrate the superiority and effectiveness of proposed framework.

Multi-View Feature Aggregation for Predicting Microbe-Disease Association.

Microbes play a crucial role in human health and disease. Figuring out the relationship between microbes and diseases leads to significant potential applications in disease treatments. It is an urgent need to devise robust and effective computational methods for identifying disease-related microbes. This work proposes a Multi-View Feature Aggregation (MVFA) scheme that integrates the linear and nonlinear features to identify disease-related microbes. We introduce a non-negative matrix tri-factorization (NMTF) model to extract linear features for diseases and microbes. Then we learn another type of linear feature by utilizing a bi-random walk model. The nonlinear feature is obtained by inputting the two kinds of linear features into a capsule neural network. These three types of features describe the associations between diseases and microbes from different views. Finally, considering the complementary of these features, we leverage a logistic regression model to combine the NMTF model predictions, bi-random walk model predictions, and the capsule neural network predictions to obtain the final microbe-disease pair scores. We apply our method to predict human microbe-disease associations on two datasets. Experimental results show that our multi-view model outperforms the state-of-the-art models in recovering missing microbe-disease associations and predicting associations for new microbes. The ablation study shows that aggregating multi-view linear and nonlinear features can improve the prediction performance. Case studies on two diseases, i.e. Type 1 diabetes and Liver cirrhosis, further validate our method effectiveness.

Safe screening rules for multi-view support vector machines

Multi-view learning aims to make use of the advantages of different views to complement each other and fully mines the potential information in the data. However, the complexity of multi-view learning algorithm is much higher than that of single view learning algorithm. Based on the optimality conditions of two classical multi-view models: SVM-2K and multi-view twin support vector machine (MvTwSVM), this paper analyzes the corresponding relationship between dual variables and samples, and derives their safe screening rules for the first time, termed as SSR-SVM-2K and SSR-MvTwSVM. It can assign or delete four groups of different dual variables in advance before solving the optimization problem, so as to greatly reduce the scale of the optimization problem and improve the solution speed. More importantly, the safe screening criterion is “safe”, that is, the solution of the reduced optimization problem is the same as that of the original problem before screening. In addition, we further give a sequence screening rule to speed up the parameter optimization process, and analyze its properties, including the similarities and differences of safe screening rules between multi-view SVMs and single-view SVMs, the computational complexity, and the relationship between the parameter interval and screening rate. Numerical experiments verify the effectiveness of the proposed methods.

Multi-View Learning to Unravel the Different Levels Underlying Hepatitis B Vaccine Response.

The immune system acts as an intricate apparatus that is dedicated to mounting a defense and ensures host survival from microbial threats. To engage this faceted immune response and provide protection against infectious diseases, vaccinations are a critical tool to be developed. However, vaccine responses are governed by levels that, when interrogated, separately only explain a fraction of the immune reaction. To address this knowledge gap, we conducted a feasibility study to determine if multi-view modeling could aid in gaining actionable insights on response markers shared across populations, capture the immune system's diversity, and disentangle confounders. We thus sought to assess this multi-view modeling capacity on the responsiveness to the Hepatitis B virus (HBV) vaccination. Seroconversion to vaccine-induced antibodies against the HBV surface antigen (anti-HBs) in early converters (n = 21; <2 months) and late converters (n = 9; <6 months) and was defined based on the anti-HBs titers (>10IU/L). The multi-view data encompassed bulk RNA-seq, CD4+ T-cell parameters (including T-cell receptor data), flow cytometry data, and clinical metadata (including age and gender). The modeling included testing single-view and multi-view joint dimensionality reductions. Multi-view joint dimensionality reduction outperformed single-view methods in terms of the area under the curve and balanced accuracy, confirming the increase in predictive power to be gained. The interpretation of these findings showed that age, gender, inflammation-related gene sets, and pre-existing vaccine-specific T-cells could be associated with vaccination responsiveness. This multi-view dimensionality reduction approach complements clinical seroconversion and all single modalities. Importantly, this modeling could identify what features could predict HBV vaccine response. This methodology could be extended to other vaccination trials to identify the key features regulating responsiveness.

MARCMV: Mining Multi-View Association Rules from Clustered Multi-Views

Data mining involves examining vast quantities of data to uncover valuable insights that can be utilized for making informed decisions and driving business objectives. The study focuses on the task of finding relationships between features belonging to two different views using multi-view model, and proposes a novel approach called MARCMV. This approach extracts multi-view association rules from different views of the same data set using multi-clustering neural model. The study finds that MARCMV outperforms conventional symbolic methods in terms of association rule quality and running time.

MVCINN: Multi-View Diabetic Retinopathy Detection Using a Deep Cross-Interaction Neural Network

Diabetic retinopathy (DR) is the main cause of irreversible blindness for working-age adults. The previous models for DR detection have difficulties in clinical application. The main reason is that most of the previous methods only use single-view data, and the single field of view (FOV) only accounts for about 13% of the FOV of the retina, resulting in the loss of most lesion features. To alleviate this problem, we propose a multi-view model for DR detection, which takes full advantage of multi-view images covering almost all of the retinal field. To be specific, we design a Cross-Interaction Self-Attention based Module (CISAM) that interfuses local features extracted from convolutional blocks with long-range global features learned from transformer blocks. Furthermore, considering the pathological association in different views, we use the feature jigsaw to assemble and learn the features of multiple views. Extensive experiments on the latest public multi-view MFIDDR dataset with 34,452 images demonstrate the superiority of our method, which performs favorably against state-of-the-art models. To the best of our knowledge, this work is the first study on the public large-scale multi-view fundus images dataset for DR detection.