← Search

Frank Permenter

2 accepted papers

2026

Much Ado About Noising: Dispelling the Myths of Generative Robotic Control

ICLR 2026poster

Generative models, like flows and diffusions, have recently emerged as popular and efficacious policy parameterizations in robotics. There has been much speculation as to the factors underlying their successes, ranging from capturing multimodal action distributions to expressing more complex behavio…

Cited by 0SourcecodeScholar
2024

Interpreting and Improving Diffusion Models from an Optimization Perspective

ICML 2024poster

Denoising is intuitively related to projection. Indeed, under the manifold hypothesis, adding random noise is approximately equivalent to orthogonal perturbation. Hence, learning to denoise is approximately learning to project. In this paper, we use this observation to interpret denoising diffusion…