MASc Thesis Published – Audio-Visual Feature Fusion through Transformers for Automated Depression Screening in Social Media Content

MASc Thesis Published

Audio-Visual Feature Fusion through Transformers for Automated Depression Screening in Social Media Content

UWSpace Repository · University of Waterloo

I am delighted to share that my MASc thesis has now been officially published in the University of Waterloo's UWSpace institutional repository. This marks an important milestone in the completion of my MASc journey and makes the full thesis openly accessible through its permanent institutional record.

The thesis presents research on transformer-based multimodal learning for automated depression screening from social media video content. It develops and evaluates methods for fusing audio and visual information, with particular emphasis on the architectures behind MDD-Net and MMFformer, and examines their effectiveness across benchmark datasets for depression detection.

I am deeply grateful to my supervisors, collaborators, and the University of Waterloo for their support throughout this work. Having the thesis archived in UWSpace provides a permanent and citable home for this stage of my research journey.