Deep learning-based multi-modal data representation learning

PhD Project details

Title: Deep learning-based multi-modal data representation learning

With the rise of heterogeneous networks like the Internet of Things, social networks, and digital networks, acquiring data from diverse sources has become more accessible. Termed “multi-modal” data, it originates from various domain sources, representing diverse data types and patterns. Effective object feature representation is vital for robust machine learning models. Given that multi-modal data encompasses comprehensive object information, multi-modal feature representation is reliable. Consequently, recent research exhibits growing interest in multi-modal representation learning, demonstrating its capacity to boost performance in subsequent machine learning tasks such as classification, clustering, and information retrieval. Nonetheless, effectively learning intra-modal and cross-modal features in multi-modal data presents challenges. In response, this research endeavours to design deep learning-based methods that effectively learn the representation of both cross-modal and intra-modal features and their interrelationships.

Supervisory team:

Prof. Richi Nayak,  Dr Md Abul Bashar