Session Outline

While the public is mostly aware of generative models for text and image generation, other domains such as 3D content and animations can highly benefit from this technology. In this talk I discuss  the challenges that we are facing such as small datasets, a lack of good metrics and high quality requirements.

Key Takeaways

  • As soon as you go beyond and images, you have a small data problem when it comes to game content
  • High quality content is hard to generate
  • It is difficult to automatically evaluate the generated content


Speaker Bio

Judith Bütepage – Senior Research Engineer | SEED – Electronic Arts

Judith Bütepage is leading the Generative AI team at SEED, Electontic Art’s internal research lab. Previously she has been a Research Scientist at Spotify. Judith holds a doctoral degree from the Royal Institute of Technology, KTH, in Machine Learning. She has been a member of the non-profit organization Stockholm AI since it’s founding in 2015. Her expertise are deep generative models applied to many domains ranging from sounds to animations and, more broadly, experience in computational neuroscience, data science, robotics and computer vision.

October 26 @ 14:10
14:10 — 14:40 (30′)

Day 2 | 26 Oct 2023 | MACHINE LEARNING + MLOPS

Judith Bütepage – Senior Research Engineer | SEED – Electronic Arts