Regularized Wasserstein Means for Aligning Distributional Data
arXiv:1812.00338
Abstract
We propose to align distributional data from the perspective of Wasserstein means. We raise the problem of regularizing Wasserstein means and propose several terms tailored to tackle different problems. Our formulation is based on the variational transportation to distribute a sparse discrete measure into the target domain. The resulting sparse representation well captures the desired property of the domain while reducing the mapping cost. We demonstrate the scalability and robustness of our method with examples in domain adaptation, point set registration, and skeleton layout.
References in corpus (9)
- Point-Set Registration: Coherent Point Drift
- Go-ICP: A Globally Optimal Solution to 3D ICP Point-Set Registration
- Adversarial Discriminative Domain Adaptation
- Fast Computation of Wasserstein Barycenters
- DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation
- Deep CORAL: Correlation Alignment for Deep Domain Adaptation
- Multilevel Clustering via Wasserstein Means
- Variational Principles for Minkowski Type Problems, Discrete Optimal Transport, and Discrete Monge-Ampere Equations
- A gradual, semi-discrete approach to generative network training via explicit Wasserstein minimization