Multimodal Subspace Support Vector Data Description

Abstract

In this paper, we propose a novel method for projecting data from multiple modalities to a new subspace optimized for one-class classification. The proposed method iteratively transforms the data from the original feature space of each modality to a new common feature space along with finding a joint compact description of data coming from all the modalities. For data in each modality, we define a separate transformation to map the data from the corresponding feature space to the new optimized subspace by exploiting the available information from the class of interest only. We also propose different regularization strategies for the proposed method and provide both linear and non-linear formulations. The proposed Multimodal Subspace Support Vector Data Description outperforms all the competing methods using data from a single modality or fusing data from all modalities in four out of five datasets.

arXiv comment26 pages manuscript (6 tables, 2 figures), 24 pages supplementary material (27 tables, 10 figures). The manuscript and supplementary material are combined as a single .pdf (50 pages) file

References in corpus (7)

• A Survey on Multi-view Learning
• One-Class Classification: Taxonomy of Study and Review of Techniques
• Generalized Multi-view Embedding for Visual Recognition and Cross-modal Retrieval
• Vehicle Tracking Using Surveillance with Multimodal Data Fusion
• Subspace Support Vector Data Description
• Analysis of PCA Algorithms in Distributed Environments
• Weighted Linear Discriminant Analysis based on Class Saliency Information

Cited by in corpus (8)

• AutoSOS: Towards Multi-UAV Systems Supporting Maritime Search and Rescue with Lightweight AI and Edge Computing
• Graph-Embedded Subspace Support Vector Data Description
• Boosting rare benthic macroinvertebrates taxa identification with one-class classification
• Linguistically-aware Attention for Reducing the Semantic-Gap in Vision-Language Tasks
• Early Myocardial Infarction Detection with One-Class Classification over Multi-view Echocardiography
• AnomalyHop: An SSL-based Image Anomaly Localization Method
• Audio-based Anomaly Detection in Industrial Machines Using Deep One-Class Support Vector Data Description
• Critical Review for One-class Classification: recent advances and the reality behind them