Abstract: A longstanding goal of unsupervised learning is to discover the true, generative factors of high dimensional data. I will discuss the relationship of this goal to what is called ""disentanglement"" in deep learning and present recent theoretical and empirical results that show that much of the recent optimism about deep learning in this context is unwarranted. At the same time, I will discuss how classical spectral methods may allow the problem to be solved under realistic conditions.
Bio: Yair Weiss is a Professor of Computer Science at the Hebrew University and the former Dean of the School of Computer Science and Engineering. His research interests include Machine Learning, Computer Vision and Neural Computation. He served as the program chair of the Neural Information Processing Systems conference (2004) and the European Conference on Computer Vision (2018). From 2004-2019 He was a Senior Fellow of the Canadian Institute for Advanced Research and he is currently a Fellow of the European Laboratory for Learning and Intelligent Systems. With his students and colleagues he has received best paper awards at UAI, NIPS, CVPR and ECCV.