Feb 11 2019 12:00 PM
Feb 11 2019 01:00 PM
-By Dr. Timo Aila (NVIDIA, Docent, Aalto
Generative methods allow a computer to automatically distill the essence of a dataset and then produce novel examples that are indistinguishable from the original data. That's the promise, but getting there has been difficult. This talk focuses on recent advances in generative adversarial networks (GAN), focusing on ideas that have finally allowed us to synthesize credible high-resolution images. I will also describe our recent work that makes the image generation more controllable by borrowing ideas from style transfer literature, and also leads to an interesting, unsupervised separation of high-level attributes (e.g. pose or identity in case of human faces) and inconsequential variation in the images (exact placement of hair, etc.).
Timo Aila works as a distinguished research scientist at NVIDIA. He has a PhD in computer graphics from Aalto University in Helsinki, Finland. He has worked extensively in real-time graphics (computer games) and offline rendering (movies such as Avatar, Tintin, Hobbit), and was also the first technical lead for the recently announced RTCore ray tracing hardware architecture. His current research focuses on machine learning and particularly on generative models with the goal of creating new powerful tools for content creation.
For more info contact Prof. Peter Wonka; firstname.lastname@example.org
Date: Monday 11th Feb 2019
Time:12:00 PM - 01:00 PM
Location: Bldg. 9 Hall 1
Refreshments: Light Lunch will be served at 11:45 AM