This special issue mainly aims to provide a platform for the related researchers and practitioners to communicate the new advancement and successful applications of multi-view learning. This special issue including 24 papers in total are carefully selected and cover a range of topics, including multi-view classification/prediction, multi-view clustering, multi-view representation learning and their applications.

The first category is multi-view classification/prediction. The paper entitled “Co-Clustering based Classification of Multi-View Data” by Syed Fawad Hussain, Mohsin Khan and Imran Siddiqi attempt to respect the consensus principle by learning shared similarity values using transfer learning and the complementary principle by learning the similarity between instances within each view supervised co-clustering measure to improve multi-view classification. The paper entitled “IBMvSVM: An Instance-Based Multi-View SVM Algorithm for Classification” by Shuang Yu, Xiongfei Li, Siru Sun, Hancheng Wang, Xiaoli Zhang and Shiping Chen design an instance-based multi-view SVM algorithm by distributing different weights to different views of an instance. The paper entitled “Multi-View Attention-Convolution Pooling Network for 3D Point Cloud Classification” by Wenju Wang, Tao Wang and Yu Cai propose a multi-view attention-convolution pooling network on multiple 2D views for 3D point cloud classification task. The paper entitled “A Multi-View Multi-Omics Model for Cancer Drug Response Prediction” by Zhijin Wang, Ziyang Wang, Yaohui Huang, Longquan Lu and Yonggang Fu present a multi-view model on multi-omics data to improve the cancer drug response prediction. The paper entitled “A Multi-Mode Traffic Flow Prediction Method with Clustering based Attention Convolution LSTM” by Xiaohui Huang, Yuming Ye, Cheng Wang, Xiaofei Yang and Liyan Xiong propose a multi-mode traffic flow prediction method with clustering-based attention convolution LSTM to model spatio-temporal data of traffic flow. The paper entitled “A Multi-View Time Series Model for Share Turnover Prediction” by Zhijin Wang, Qiankun Su, Guoqing Chao, Bing Cai, Yaohui Huang and Yonggang Fu propose a multi-view time series model to capture temporal dynamics to predict share turnover values more accurately.

The second category concerns multi-view clustering. The paper entitled “Multi-View K-Proximal Plane Clustering” by Feixiang Sun, Xijiong Xie, Jiangbo Qian, Yu Xin, Yuqi Li, Chong Wang and Guoqing Chao present a multi-view k-proximal plane clustering to deal with data points that are clustered along a straight line. The paper entitled “Semi-Supervised Multi-View Binary Learning for Large-Scale Image Clustering” by Mingyang Liu, Zuyuan Yang, Wei Han, Junhang Chen and Weijun Sun design a multi-view binary learning method to deal with large-scale clustering problem in semi-supervised way. The paper entitled “Nonconvex Low-Rank and Sparse Tensor Representation for Multi-View Subspace Clustering” by Shuqin Wang, Yongyong Chen, Yigang Cen, Linna Zhang, Hengyou Wang and Viacheslav Voronin design a novel multi-view subspace clustering model by forming a nonconvex optimization problem involving low-rank and sparse tensor representation. The paper entitled “Incomplete Multi-View Clustering with Multiple Imputation and Ensemble Clustering” by Guoqing Chao, Songtao Wang, Shiming Yang, Chunshan Li and Dianhui Chu propose a novel multi-view clustering model to deal with any missing value case with multiple imputation and ensemble clustering. The paper entitled “Efficient Multi-view Clustering Networks” by Guanzhou Ke, Zhiyong Hong, Wenhua Yu, Xin Zhang and Zeyi Liu aim at promoting the efficiency of multi-view clustering algorithms by designing an alternating process of an approximation and an instruction. The paper entitled “Incomplete Multi-view Clustering based on Weighted Sparse and Low Rank Representation” by Liang Zhao, Jie Zhang, Tao Yang and Zhikui Chen propose an incomplete multi-view clustering algorithm by leveraging subspace learning with double constraints to capture global and local data relationships and adopting a weighting mechanism to reduce the negative impact of missing data. The paper entitled “Robust Deep Multi-View Subspace Clustering Networks with a Correntropy-induced Metric” by Xiaomeng Si, Qiyue Yin, Xiaojie Zhao and Li Yao design a position-aware diversity regularization to make full use of the complementary information inside multi-view data, a correntropy-induced metric to handle complex noise distributions to improve the robustness of deep multi-view subspace clustering networks. The paper entitled “Incomplete Multi-View Clustering with Incomplete Graph-Regularized Orthogonal Non-Negative Matrix Factorization” by Naiyao Liang, Zuyuan Yang, Zhenni Li and Wei Han design an incomplete multi-view clustering algorithm by employing the geometric structure and orthogonal nonnegative matrix factorization.

The third category is about multi-view representation learning. The paper entitled “Deep Mutual Information Multi-View Representation for Visual Recognition” by Xianfa Xu, Zhe Chen and Fuliang Yin propose an anto-encoder network which maximizes the mutual information between the latent representation and the original feature and maximizes the canonical correlation of different view mean vectors to learn a robust representation in visual recognition task. The paper entitled “Trace Ratio Criterion for Multi-view Discriminant Analysis” by Mei Shi, Zhihui Li, Xiaowei Zhao, Pengfei Xu, Baoying Liu and Jun Guo design a trace ratio criterion for multi-view representation learning by taking the intra-view and inter-view correlation across multiple views into consideration.

The fourth category covers multi-view learning applications. The paper entitled “Edge-Enhanced Dual Discriminator Generative Adversarial Network for Fast MRI with Parallel Imaging Using Multi-View Information” by Jiahao Huang, Weiping Ding, Jun Lv, Jingwen Yang, Hao Dong, Javier De Ser, Jun Xia, Tiaojuan Ren, Stephen T. Wong and Guang Yang design a new Generative Adversarial Network framework to make full use of multi-view information to enhance the edge information to reconstruct the MRI, thus MRI can be obtained fast to assist the diagnosis. The paper entitled “Online Unsupervised Cross-View Discrete Hashing for Large-Scale Retrieval” by Xuan Li, Wei Wu, Yun-Hao Yuan, Shirui Pan and Xiaobo Shen design an online unsupervised cross-view discrete hashing that considers similarity among newly arriving data and old data that is absent by selecting a few anchors. The paper entitled “M-FFN: Multi-Scale Feature Fusion Network for Image Captioning” by Jeripothula Prudviraj, Chalavadi Vishnu and C. Krishna Mohan present a feature fusion network for image captioning task to incorporate discriminative features and scene contextual information of an image. The paper entitled “Sign Language Recognition and Translation Network based on Multi-View Data” by Roghui Li and Lu Meng propose a transformer and graph convolutional network based multi-view model for sign language recognition and translation. The paper entitled “Speech Synthesis with Face Embeddings” by Xing Wu, Sihui Ji, Jianjia Wang and Yike Guo aim to synthesize speech from the same person’s image by designing a model consisting of voice encoder, face encoder and an improved multi-speaker text-to-speech engine. The paper entitled “Cross-View Vehicle Re-identification Based on Graph Matching” by Chao Zhang, Chule Yang, Dayan Wu, Hongbin Dong and Baosong Deng propose a systematic framework to realize the alignment and discrimination of key features from two views by learning high-order relationships and topological information. The paper entitled “Locality Sensitive Hashing with Bit Selection” by Wenhua Zhou, Huawen Liu, Jungang Lou and Xin Chen present a new locality sensitive hashing algorithm by selecting the bits with high importance and less similarity to assist nearest neighbor search in image retrieval field. The paper entitled “Face Aging with Pixel-Level Alignment GAN” by Xing Wu, Yafei Zhang, Qing Li, Yangyang Qi, Jianjia Wang and Yike Guo design a face aging model consisting of age estimation, identity preservation and image de-noising.

Due to its comprehensiveness and robust nature in describing objects, multi-view learning is one of the fundamental tasks of the machine learning community to solve real-world learning problems. With 24 selected papers in the field, this special issue provides a focused discussion about recent advancement in multi-view learning.

Finally, guest editors would like to cordially thank all the authors for their supports and contributions. Our thanks also extend to all the dedicated reviewers assisting in the papers’ evaluation and selection.