Domain Adaptation Optimal Transport

University jean monnet domain adaptation using optimal transport.

Domain adaptation optimal transport. If the adaptation is done correctly models built on a specific data representation become more robust when confronted to data depicting the same classes but described by another observation system. Here the target has the particularity to present new classes not present in the source domain. 0 share.

Optimal transport for domain adaptation. Leo g autheron supervisors. Index terms unsupervised domain adaptation optimal t ranspor t t ransfer learning visual adaptation classification.

We present a 2 step optimal transport approach that performs a mapping from a source distribution to a target distribution. Can use estimated mapping perrot et al 2016 seguy et al 2017. Ot however is very sensitive to outliers samples with large noise in the data since in its objective function every sample including outliers is weighed similarly due to the marginal constraints.

Carole lartizien cnrs researcher ievgen redko associate professor a thesis submitted in partial fulfillment of the machine learning and. The first step of the approach aims at rejecting the samples issued from these new classes using an optimal transport plan. Ieee transactions on pattern analysis and machine intelligence institute of electrical and electronics engineers 2016 39 9 pp 1853 1865.

Optimal transport for domain adaptation. Optimal transport ot distances such as wasserstein have been used in several areas such as gans and domain adaptation. Application to prostate cancer mapping author.

Domain adaptation is one of the most challenging tasks of modern data analytics. The second step solves the target class ratio. Optimal transport for domain adaptation dataset class 1 class 2 samples samples classifier on optimal transport samples samples classification on transported samples classifier on discussion works very well in practice for large class of transformation courty et al 2016a.