Multi-level Feature Learning for Contrastive Multi-view Clustering

Xu, Jie; Tang, Huayi; Ren, Yazhou; Peng, Liang; Zhu, Xiaofeng; He, Lifang

doi:10.1109/cvpr52688.2022.01558

Cited by 95 publications

(54 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, we collect nine state-of-the-art multi-view clustering methods and compare them with the proposed IMVC. 1) AMGL [38] proposed a parameter-free view weights learning scheme to distinguish the importance of various views; 2) CSMSC [39] explored the consistent subspace representation and viewspecific representations; 3) GMC [40] learned a global graph matrix with a certain number of connected components; 4) CGL [41] recovered a low-rank tensor space, wherein the spectral embedding of each view are optimized; 5) EOMSC-CA [42] proposed a unified framework with simultaneously performing anchor selection and graph construction; 6) MvD-SCN [43] utilized the uniformity and diversity networks to learn the consistent and view-specific representations; 7) DSRL [44] developed an deep sparse regularizer learning method for multi-view clustering; 8) DMSC-UDL [45] mined the consensus subspace representation combining the diverse information of different views; 9) MFLVC [46] used the data features under different views as the contrastive entities and performed the contrastive learning to reinforce the sample discrimination.…”

Section: B Baselines and Evaluation Metrics 1) Baseline Methodsmentioning

confidence: 99%

Low-rank tensor approximation with local structure for multi-view intrinsic subspace clustering

Yang

Chen

et al. 2022

Information Sciences

View full text Add to dashboard Cite

Section: B Baselines and Evaluation Metrics 1) Baseline Methodsmentioning

confidence: 99%

Low-rank tensor approximation with local structure for multi-view intrinsic subspace clustering

Yang

Chen

et al. 2022

Information Sciences

View full text Add to dashboard Cite

“…Especially in the field of computer vision, contrastive learning methods have produced excellent results [23]. As an example, methods such as SimClR or MoCo [17, 22, 24] minimise the InfoNCE loss function [25] to maximise the upper bound of mutual information. Because dealing with negative samples is inconvenient, later contrastive learning algorithms [26, 27] have successfully replaced the contrast task with the prediction task without the need for negative samples.…”

Section: Related Workmentioning

confidence: 99%

“…Almost all existing contrastive learning methods [17, 22, 24, 26, 27] are designed to handle single‐view data, exhaustively exploring various data augmentations to build different views/augments. To the best of our knowledge, this is the first time double contrastive learning has been used on the IMC problem to handle the challenge of mutual promotion of consistency learning and data recovery from a different perspective.…”

Section: Related Workmentioning

confidence: 99%

“…Actually, due to the variety and diversity of incomplete data from multiple views, it is quite challenging to utilise incomplete data to its full potential which makes the model more robust in severe scenarios with high missing rates. To make up for the lack of complete data pairs and comprehensively utilise the hidden information in incomplete data, contrastive learning has been applied in IMC [14–18]. However, while learning consistency from different views, contrastive learning in the feature space will result in the loss of complementary information.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Self‐supervised image clustering from multiple incomplete views via constrastive complementary generation

Wang

Yang³

et al. 2022

IET Computer Vision

View full text Add to dashboard Cite

Incomplete Multi‐View Clustering aims to enhance clustering performance by using data from multiple modalities. Despite the fact that several approaches for studying this issue have been proposed, the following drawbacks still persist: (1) It is difficult to learn latent representations that account for complementarity yet consistency without using label information; (2) and thus fails to take full advantage of the hidden information in incomplete data results in suboptimal clustering performance when complete data is scarce. In this study, Contrastive Incomplete Multi‐View Image Clustering with Generative Adversarial Networks (CIMIC‐GAN), which uses Generative Adversarial Network (GAN) to fill in incomplete data and uses double contrastive learning to learn consistency on complete and incomplete data is proposed. More specifically, considering diversity and complementary information among multiple modalities, we incorporate autoencoding representation of complete and incomplete data into double contrastive learning to achieve learning consistency. Integrating GANs into the autoencoding process can not only take full advantage of new features of incomplete data, but also better generalise the model in the presence of high data missing rates. Experiments conducted on four extensively used data sets show that CIMIC‐GAN outperforms state‐of‐the‐art incomplete multi‐View clustering methods.

show abstract

“…These would cause suboptimal performance since they may learn meaningless modality-private noise information of each modality, introducing additional noises to domain clustering. Yet, this can be avoided by applying cross-modal contrastive learning, which is able to learn the common semantics across all modalities while reducing their meaningless modality-private noise information [33].…”

Section: Introductionmentioning

confidence: 99%

Deciphering Spatial Domains by Integrating Histopathological Image and Transcriptomics via Contrastive Learning

Zeng

Yin

Luo

et al. 2022

Preprint

View full text Add to dashboard Cite

Recent advances in spatial transcriptomics have enabled measurements of gene expression at cell/spot resolution meanwhile retaining both the spatial information and the histopathological images of the tissues. Deciphering the spatial domains of spots in the tissues is a vital step for various downstream tasks in spatial transcriptomics analysis. Existing methods have been developed for this purpose by combining gene expression and histopathological images to conquer noises in gene expression. However, current methods only use the histopathological images to construct spot relations without updating in the training stage, or simply concatenate the information from gene expression and images into one feature vector. Here, we propose a novel method ConGI to accurately decipher spatial domains by integrating gene expression and histopathological images, where the gene expression is adapted to image information through contrastive learning. We introduce three contrastive loss functions within and between modalities to learn the common semantic representations across all modalities while avoiding their meaningless modality-private noise information. The learned representations are then used for deciphering spatial domains through a clustering method. By comprehensive tests on tumor and normal spatial transcriptomics datasets, ConGI was shown to outperform existing methods in terms of spatial domain identification. More importantly, the learned representations from our model have also been used efficiently for various downstream tasks, including trajectory inference, clustering, and visualization.

show abstract

Multi-level Feature Learning for Contrastive Multi-view Clustering

Cited by 95 publications

References 33 publications

Low-rank tensor approximation with local structure for multi-view intrinsic subspace clustering

Low-rank tensor approximation with local structure for multi-view intrinsic subspace clustering

Self‐supervised image clustering from multiple incomplete views via constrastive complementary generation

Deciphering Spatial Domains by Integrating Histopathological Image and Transcriptomics via Contrastive Learning

Contact Info

Product

Resources

About