A geometric view of optimal transportation and generative model
From MaRDI portal
Publication:669014
DOI10.1016/J.CAGD.2018.10.005zbMATH Open1430.49050arXiv1710.05488OpenAlexW2963875852MaRDI QIDQ669014FDOQ669014
Li Cui, David Xianfeng Gu, Kehua Su, Shing-Tung Yau, Na Lei
Publication date: 20 March 2019
Published in: Computer Aided Geometric Design (Search for Journal in Brave)
Abstract: In this work, we show the intrinsic relations between optimal transportation and convex geometry, especially the variational approach to solve Alexandrov problem: constructing a convex polytope with prescribed face normals and volumes. This leads to a geometric interpretation to generative models, and leads to a novel framework for generative models. By using the optimal transportation view of GAN model, we show that the discriminator computes the Kantorovich potential, the generator calculates the transportation map. For a large class of transportation costs, the Kantorovich potential can give the optimal transportation map by a close-form formula. Therefore, it is sufficient to solely optimize the discriminator. This shows the adversarial competition can be avoided, and the computational architecture can be simplified. Preliminary experimental results show the geometric method outperforms WGAN for approximating probability measures with multiple clusters in low dimensional space.
Full work available at URL: https://arxiv.org/abs/1710.05488
Cites Work
- Polar factorization and monotone rearrangement of vector‐valued functions
- Title not available (Why is that?)
- Title not available (Why is that?)
- Variational principles for Minkowski type problems, discrete optimal transport, and discrete Monge-Ampère equations
- Convex Polyhedra
- A discrete uniformization theorem for polyhedral surfaces
- Semidual Regularized Optimal Transport
- From Knothe's Rearrangement to Brenier's Optimal Transport Map
Cited In (16)
- Multi-projection of unequal dimension optimal transport theory for generative adversary networks
- Randomized methods for computing optimal transport without regularization and their convergence analysis
- Discrete Lie flow: a measure controllable parameterization method
- A structural model on a hypercube represented by optimal transport
- Connecting GANs, mean-field games, and optimal transport
- Ellipsoidal conformal and area-/volume-preserving parameterizations and associated optimal mass transportations
- Projection-based techniques for high-dimensional optimal transport problems
- Radon-Sobolev variational auto-encoders
- A Scalable Deep Learning Approach for Solving High-Dimensional Dynamic Optimal Transport
- Convergence Analysis of Volumetric Stretch Energy Minimization and Its Associated Optimal Mass Transport
- The Singularity Set of Optimal Transportation Maps
- Optimal transport for generative models
- Projected gradient method combined with homotopy techniques for volume-measure-preserving optimal mass transportation problems
- The singularity set of optimal transportation maps
- Optimal transport for generative models
- Geometrical Insights for Implicit Generative Modeling
Uses Software
This page was built for publication: A geometric view of optimal transportation and generative model
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q669014)
