I’m a Ph.D. student at the University of Toronto studying machine learning and artificial intelligence under the supervision of Richard Zemel.¬†While here, I have also had the privilege to work with Geoffrey Hinton, Brendan Frey and Ruslan Salakhutdinov. Previously, I completed a masters degree at the University of British Columbia under the supervision of Nando de Freitas (now at Oxford). Here’s my CV.

I try to maintain a broad set of interests within machine learning, although I specialize in neural networks. Most recently, I have become interested in Bayesian optimization for automatically tuning the meta-parameters of neural networks (and other machine learning models). This approach has already had tremendous success on several benchmark datasets and often out-performs experts in tuning these models. I think this is a promising avenue for making deep learning more accessible, both within the machine learning community and beyond.

In the past I’ve worked on graphical models, inference algorithms, Bayesian methods, time-series problems and unsupervised feature learning. My main focus has been applying a probabilistic perspective to machine learning methods. This approach often provides a principled way to analyze existing techniques while serving as a launching point for interesting new extensions.

Some applications I’ve worked on in the past include: object recognition, speech recognition, environmental monitoring, and ranking for web search.

Latest Publications

Learning Unbiased Features
Yujia Li, Kevin Swersky and Richard Zemel
Neural Information Processing Systems Workshop on Transfer and Multi-Task Learning