Presentation
ComboStoc: Combinatorial Stochasticity for Diffusion Generative Models
SessionImage Generation
DescriptionWe study an important factor in diffusion generative models: combinatorial complexity. Existing training schemes insufficiently sample the joint space of dimensions and attributes, leading to weaker test time performance. We address this with ComboStoc, a simple stochastic process that better exploits underlying combinatorial structure.
Event Type
Technical Paper
TimeWednesday, 22 July 20269:50am - 10:00am PDT
LocationRoom 408 B
Digital Library
PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Wednesday, 22 July 20269:00am - 10:30am PDTRoom 408 B
Artificial Intelligence/Machine Learning
Generative AI
Modeling
Full Conference Supporter
Full Conference
