Wild Variational Approximations

How researchers see this paper

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

Adversarial Variational Bayes: Unifying Variational Autoencoders and Generative Adversarial Networks (2017, methodology): “The idea to use adversarial training to improve the encoder network was also suggested by Goodfellow in an exploratory talk he gave at NIPS 2016 and by Li & Liu (2016).”
Advances in Variational Inference (2017, methodology): “In this case, one can employ a Generative Adversarial Networks (GAN) [54] style discriminator T that discriminates the prior from the variational distribution, T ðx; zÞ 1⁄4 log qfðzjxÞ log pðzÞ [102], [118].”
Variational Implicit Processes (2018, methodology): “In approximate inference context, implicit distributions have also been deployed to postulate ﬂexible approximate posterior distributions [56, 58, 60, 61, 76, 87, 90].”
Gradient Estimators for Implicit Models (2017, methodology): “A related methodology, wild variational inference [22, 19] assumes a tractable joint density p(x, z) but uses implicit proposal distributions to approximate an intractable exact posterior p(z|x).”
Debiasing Evidence Approximations: On Importance-weighted Autoencoders and Jackknife Variational Inference (2018, background): “…2016) for variational inference; this line of work includes adversarial variational Bayes (Mescheder et al., 2017), wild variational inference (Li & Liu, 2016), deep implicit models (Tran et al., 2017), implicit variational models (Huszár, 2017), and adversarial message passing…”