Sinkhorn-AutoDiff: Tractable Wasserstein Learning of Generative Models

By Aude Genevay, G. Peyr'e, Marco Cuturi. 2017

48 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Parametric Adversarial Divergences are Good Task Losses for Generative Modeling (2017, methodology): “We implement the Sinkhorn-AutoDiff algorithm (Genevay et al., 2017) to compute the entropy-regularized L2-Wasserstein distance between minibatches of training images and generated images.”
A Fast Proximal Point Method for Computing Wasserstein Distance (2018, methodology): “In the tests, we observe the method in [14] suffers from shrinkage problem, i.e. the generated distribution tends to shrink towards the target mean.”
DeepJDOT: Deep Joint distribution optimal transport for unsupervised domain adaptation (2018, methodology): “Therefore, in this work we propose to solve the problem with a stochastic approximation using minibatches from both the source and target domains [35].”
Improving GANs Using Optimal Transport (2018, methodology): “Genevay et al. (2017b) then approximate this distance by evaluating it on mini-batches of data X , Y consisting of K data vectors x , y .”
Prescribed Generative Adversarial Networks (2019, methodology): “…and optimization methods (see, e.g., Li et al., 2015; Dziugaite et al., 2015; Nowozin et al., 2016; Arjovsky et al., 2017; Ravuri et al., 2018; Genevay et al., 2017) or using ideas from VAE s (Makhzani et al., 2015; Mescheder et al., 2017; Dumoulin et al., 2016; Donahue et al., 2016;…”
GAN and VAE from an Optimal Transport Point of View (2017, background): “Following [4, 7], in the special case of a generative model µ θ = g θ(cid:93) ζ , formula (1) can be conveniently re-written as This is advantageous because now π is deﬁned over Z × X , which is lower-dimensional than X × X , and also because, as in Equation (2), θ does not appear in the…”
Learning Latent Permutations with Gumbel-Sinkhorn Networks (2018, methodology): “Finally, we note our work shares with Salimans et al. (2018); Genevay et al. (2017) in that the OT cost function (here, the matrix X ) is learned using an artiﬁcial neural network.”
A Fast Proximal Point Method for Computing Exact Wasserstein Distance (2018, methodology): “Can we just employ Sinkhorn distance with a moderately sized for machine learning problems so that we can get the benefits of the reduced complexity? Some applications show Sinkhorn distance can generate good results with a moderately sized [15, 20].”