Alex N. Wang

I’m a third year Ph.D. student in the CILVR lab advised by Mengye Ren at New York University. My research is supported by the NSERC PGS-D scholarship. I obtained my M.Sc. in Computer Science advised by Richard Zemel and B.A.Sc in Engineering Science from the University of Toronto.

My research is focused on vision-centric machine learning. In particular, I am interested in:

• Self-supervised representation learning
  How single-subject (iconic) datasets like ImageNet have influenced architectures, augmentation techniques, learned representations and ultimately the ability to learn from dense, real-world image and video data?
• Impact on on large-scale vision
  How these iconic representations and training methods influence large-scale multimodal and vision models? Alternatively, can a strong semantic model like DINOv2 support the training of more general, generative vision models?
• Learning from synthetic data
  How synthetic data can contribute to visual understanding, particularly the distinction between learning effective visual features vs. real-world concepts and classes using generated images?

Looking ahead, I believer there may be inherent limitations to SSL methods stemming from their development around ImageNet data. However, I believe these models and their representations will serve as a strong foundation upon which to build large-scale generative vision models.

Feel free to contact me at anw2067 [at] cims.nyu.edu