Abstract: The ultimate goal of generative learning is to learn a model of how data is generated in the real world. Such models allow us to generate novel samples from a data distribution and have applications in (i) content generation (e.g., image, video, speech, text, music, and molecule generation) and (ii) representation learning and semi-supervised training (e.g., learning from limited labeled data). In this talk, I will first briefly review different classes of deep generative models, and then, I will focus on the recent developments in this field including the state-of-the-art variational auto-encoders (VAEs), denoising diffusion models, and generative adversarial networks (GANs). This talk will mostly focus on generative models designed for image synthesis, however, the techniques discussed in the talk can be easily applied to other data types.
Bio: Arash Vahdat is a senior research scientist at NVIDIA research specializing in machine learning and computer vision. Before joining NVIDIA, he was a research scientist at D-Wave Systems where he worked on deep generative learning and weakly supervised learning. Prior to D-Wave, Arash was a research faculty member at Simon Fraser University (SFU), where he led research on deep video analysis and taught graduate-level courses on big data analysis. Arash obtained his Ph.D. and MSc from SFU under Greg Mori’s supervision working on latent variable frameworks for visual analysis. His current areas of research include deep generative learning, weakly supervised learning, efficient neural networks, and probabilistic deep learning.