Qingjun Jiang

Understanding Multimodal Learning From Modality Fusion and Alignment Perspectives.

This paper develops a dynamic strategy that jointly optimizes modality fusion and alignment, and develops a learning-based strategy using a bi-level optimization framework and theoretically proves the convergence of the learning algorithm to ensure its reliability.