Information Fusion: Scaling Subspace-Driven Approaches

TL;DR

This paper introduces DRoGSuRe, a deep learning method that leverages subspace structures in multimodal data for robust clustering, outperforming existing approaches especially under noisy conditions.

Contribution

The paper proposes a novel deep multimodal clustering approach that exploits group subspace structures and demonstrates improved robustness over state-of-the-art methods.

Findings

DRoGSuRe outperforms DMSC in noisy environments.

The approach effectively captures the deep structure of multimodal data.

Experimental results show competitive performance across datasets.

Abstract

In this work, we seek to exploit the deep structure of multi-modal data to robustly exploit the group subspace distribution of the information using the Convolutional Neural Network (CNN) formalism. Upon unfolding the set of subspaces constituting each data modality, and learning their corresponding encoders, an optimized integration of the generated inherent information is carried out to yield a characterization of various classes. Referred to as deep Multimodal Robust Group Subspace Clustering (DRoGSuRe), this approach is compared against the independently developed state-of-the-art approach named Deep Multimodal Subspace Clustering (DMSC). Experiments on different multimodal datasets show that our approach is competitive and more robust in the presence of noise.