Skip to main content

Showing 1–2 of 2 results for author: Kornhardt, G

Searching in archive cs. Search in all archives.
.
  1. arXiv:2212.00133  [pdf, ps, other

    cs.LG math.OC stat.ML

    Universal Neural Optimal Transport

    Authors: Jonathan Geuter, Gregor Kornhardt, Ingimar Tomasson, Vaios Laschos

    Abstract: Optimal Transport (OT) problems are a cornerstone of many applications, but solving them is computationally expensive. To address this problem, we propose UNOT (Universal Neural Optimal Transport), a novel framework capable of accurately predicting (entropic) OT distances and plans between discrete measures for a given cost function. UNOT builds on Fourier Neural Operators, a universal class of ne… ▽ More

    Submitted 12 June, 2025; v1 submitted 30 November, 2022; originally announced December 2022.

    Comments: 37 pages, 19 figures, accepted to ICML 2025

    MSC Class: 68T07 (Primary) 90C08 (Secondary) ACM Class: I.2.6; G.3; G.4

  2. arXiv:2109.01105  [pdf, other

    cs.LG cs.IT eess.SP stat.ML

    Solving Inverse Problems with Conditional-GAN Prior via Fast Network-Projected Gradient Descent

    Authors: Muhammad Fadli Damara, Gregor Kornhardt, Peter Jung

    Abstract: The projected gradient descent (PGD) method has shown to be effective in recovering compressed signals described in a data-driven way by a generative model, i.e., a generator which has learned the data distribution. Further reconstruction improvements for such inverse problems can be achieved by conditioning the generator on the measurement. The boundary equilibrium generative adversarial network… ▽ More

    Submitted 2 September, 2021; originally announced September 2021.

    Comments: 13 pages, 8 figures