Text Representation through Multimodal Variational Autoencoder for One-Class Learning

Abstract

Multi-class learning (MCL) methods perform Automatic Text Classification (ATC), which requires labeling for all classes. MCL fails when there is no well-defined information about the classes and requires a great effort to label instances. One-Class Learning (OCL) can mitigate these limitations since the training only has instances from one class, reducing the labeling effort and making the ATC more appropriate for open-domain applications. However, OCL is more challenging due to the lack of counterexamples for model training, requiring more robust representations. However, most studies use unimodal representations, even though different domains contain other information that can be used as modalities. Thus, this study proposes the Multimodal Variational Autoencoder (MVAE) for OCL. MVAE is a multimodal method that learns a new representation from more than one modality, capturing the characteristics of the interest class in an adequate way. MVAE explores semantic, density, linguistic, and spatial information modalities. The main contributions are: (i) a multimodal method for ATC through OCL; (ii) MVAE for fake news detection; (iii) relevant reviews detection via MVAE; and (iv) sensing events through MVAE.