Multi-view Representation Learning and Multi-modal Information Representation

Motivation and Topics

During the recent decades, with the rapid development of information and computer technology, many fields have transformed data-poor areas to increasingly data-rich fields of research. Meanwhile, huge amount of data are often collected and extracted from multiple information sources and observed from various views. For example, a person can be identified by fingerprint, face, signature or iris with information obtained from multiple sources; an object can also be represented as multi-views, which can be seen as different feature subsets of the image; the news can be reported by a combination of texts, images, and videos on the internet; More and more information is represented by multi-view or multi modal data. To overcome the limitations of a single-view or single-modal data representation, different views and modals can be leveraged to provide complementary information to each other, and comprehensively characterize the data. Thus, multi-view representation learning and multi-modal information representation have raised widespread concerns in diverse applications. The main challenge is how to effectively explore the consistency and complementary properties from different views and modals for improving the multi-view learning performance.

The goal of this special issue in Pattern Recognition Letters is to collect high-quality articles focusing on developments, trends, and research solutions of multi-view representation learning and multi-modal information representation in range of applications. The topics of interest include, but are not limited to:

Ø Feature learning techniques (feature selection/reduction/fusion, subspace learning, sparse coding, etc.) for multi-view data. 

Ø Multi-view data based real-world applications, e.g., object detection/tracking, image segmentation, video understanding/categorization, scene understanding, action recognition, classification/clustering tasks, etc. 

Ø Advanced deep Learning techniques for multi-view data learning and understanding. 

Ø Structured/semi-structured multi-view data learning (e.g., one-shot learning, zero-shot learning, supervised learning, and semi-/unsupervised learning). 

Ø Multi-view missing data completion.

Ø Multi-modal information retrieval and classification.

Ø Large-scale multi-view data learning and understanding. 

Ø Multi-task/Transfer learning for multi-view data understanding.

Ø Multi-modal data based medical applications (diagnosis, reconstruction, segmentation, registration, etc.) 

Ø Multi-modal data based medical image analysis with advanced deep learning techniques.

Ø Multi-modal data based remote sensing image analysis. 

Ø Survey papers with regards to topics of multi-view representation learning and understanding. 

Ø New benchmark datasets collection for multi-view data learning.

Important Dates

Ø Submission period: March 1~31, 2021

Ø First review notification: June 1, 2021

Ø Revision submission deadline: August 1, 2021

Ø Second review notification: September 15, 2021

Ø Acceptance notification: October 15, 2021

Ø Date at which ALL papers are expected to be in final status: November 1, 2021

Submission Guidelines

Authors should prepare their manuscripts according to the online submission requirements of Pattern Recognition Letters at https://ees.elsevier.com/prletters/ during the submissions period. When uploading the papers through the online system, Authors should select “VSI: MVRLMMIR” as article type. Please submit your manuscript before the submission deadline.

Papers should be prepared by adhering to the PRLetters guidelines by taking into account that VSI papers follow the same submission rules as regular articles. The review process will follow the standard PRLetters scheme, and each paper will be reviewed by (at least) 2 referees. In general, only two reviewing rounds will be possible, out of which major revision is possible only for the first round. Authors should know that their papers will be most possibly rejected if after the 2nd reviewing round still need major revision.

Authors should know that the submitted papers should not have been published previously, nor be under consideration for publication elsewhere. If the submissions are extended works of previously published papers, the original works should be quoted in the References and a description of the changes that have been made should be provided. The submissions should clearly demonstrate the evidence of benefits to society or large communities. Originality and impact on society, in combination with the media nature and innovative technical aspects of the proposed solutions, will be the major evaluation criteria.

Guest Editors

MGE – Dan Song

Assistant Professor – Tianjin University (China)

dan.song@tju.edu.cn

GE – Wenshu Zhang

Assistant Professor – Cardiff Metropolitan University (UK)

WZhang@cardiffmet.ac.uk

GE – Tongwei Ren

Associate Professor – Nanjing University (China)

rentw@nju.edu.cn

GE – Xiaojun Chang

Professor – Monash University (Australia)

Xiaojun.Chang@monash.edu