Absent Multiview Semisupervised Classification

Zhuge, Wenzhang; Luo, Tingjin; Fan, Ruidong; Tao, Hong; Hou, Chenping; Yi, Dongyun

doi:10.1109/tcyb.2023.3241171

Cited by 5 publications

(3 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To validate the effectiveness of DIMvLN, we compare it with six state-of-the-art approaches, which can be categorized into two groups: traditional methods and deep methods. Traditional methods include SLIM (Yang et al 2018), AMSC (Zhuge et al 2023), AMMSS (Cai et al 2013), while deep methods include CPM-Nets (Zhang et al 2019), DCP (Lin et al 2022) and TMC (Han et al 2023). Similar to (Zhuge et al 2023), we introduce a grid search strategy to determine the optimal parameter within the recommended range for all compared 1 https://cvml.ista.ac.at/AwA/ methods and adopt the recommended network structures as their baselines.…”

Section: Training Strategymentioning

confidence: 99%

“…Traditional methods include SLIM (Yang et al 2018), AMSC (Zhuge et al 2023), AMMSS (Cai et al 2013), while deep methods include CPM-Nets (Zhang et al 2019), DCP (Lin et al 2022) and TMC (Han et al 2023). Similar to (Zhuge et al 2023), we introduce a grid search strategy to determine the optimal parameter within the recommended range for all compared 1 https://cvml.ista.ac.at/AwA/ methods and adopt the recommended network structures as their baselines. Note that CPM-Nets, DCP and TMC are supervised methods.…”

Section: Training Strategymentioning

confidence: 99%

“…The Thirty-Eighth AAAI Conference on Artificial Intelligence ties (SLIM) by learning view-specific projection and a common label matrix simultaneously based on the reconstruction of the incomplete similarity. Besides, Zhuge et al (2023) proposed the absent multi-view semi-supervised classification (AMSC) method to jointly learn the view-specific and shared label indicator matrices, and classify the unlabeled data through a label propagation mechanism by combining the extra-view and intra-view similarity loss. However, both SLIM and AMSC employ traditional shallow methods to learn the information of features and labels, which are difficult to capture high-level semantic information hidden in all views.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep Incomplete Multi-View Learning Network with Insufficient Label Information

Jiang,

Luo,

Liang

2024

AAAI

View full text Add to dashboard Cite

Due to the efficiency of integrating semantic consensus and complementary information across different views, multi-view classification methods have attracted much attention in recent years. However, multi-view data often suffers from both the miss of view features and insufficient label information, which significantly decrease the performance of traditional multi-view classification methods in practice. Learning for such simultaneous lack of feature and label is crucial but rarely studied. To tackle these problems, we propose a novel Deep Incomplete Multi-view Learning Network (DIMvLN) by incorporating graph networks and semi-supervised learning in this paper. Specifically, DIMvLN firstly designs the deep graph networks to effectively recover missing data with assigning pseudo-labels of large amounts of unlabeled instances and refine the incomplete feature information. Meanwhile, to enhance the label information, a novel pseudo-label generation strategy with the similarity constraints of unlabeled instances is proposed to exploit additional supervisory information and guide the completion module to preserve more semantic information of absent multi-view data. Besides, we design view-specific representation extractors with the autoencoder structure and contrastive loss to learn high-level semantic representations for each view, promote cross-view consistencies and augment the separability between different categories. Finally, extensive experimental results demonstrate the effectiveness of our DIMvLN, attaining noteworthy performance improvements compared to state-of-the-art competitors on several public benchmark datasets. Code will be available at GitHub.

show abstract