[1] Explainability and Trustworthiness in Deep Learning;

[2] Multi-modal Representation Learning (Video Content Understanding, Multimodal Large Language Model);

[3] Causal Representation Learning.