Structured VAEs: Composing Probabilistic Graphical Models and Variational Autoencoders

By Matthew J. Johnson, David Kristjanson Duvenaud, Alexander B. Wiltschko, Sandeep R. Datta, Ryan P. Adams. 2016

27 citations recorded by the source corpus.

Subfields: Generative Models, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Generative Models

1. Yoshua Bengio
2. Aaron C. Courville
3. Stephan Mandt
4. Yisong Yue
5. Frank D. Wood
6. Iulian Serban
7. Tuan Anh Le
8. Jan-Willem van de Meent

In Probabilistic & Bayesian Methods

1. Frank D. Wood
2. Hongseok Yang
3. Jan-Willem van de Meent
4. Brooks Paige
5. Yoshua Bengio
6. Tuan Anh Le
7. Atilim Gunes Baydin
8. Aaron C. Courville

Citations in the corpus

• An Introduction to Probabilistic Programming (2018, background): “Some work has been done to increase the generality of the modeling language, such as making the decoder generative model a graphical model (Johnson et al., 2016).”
• Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data (2016, background): “The third model [9] does not violate Eq.”
• Inference Compilation and Universal Probabilistic Programming (2016, background): “Approaches in this camp generally produce recognition networks that nonlinearly transform observational data at test time into parameters of a variational posterior approximation, albeit one with less conditional structure, excepting the recent work of Johnson et al. (2016).”
• Factorized Variational Autoencoders for Modeling Audience Reactions to Movies (2017, background): “[17] combine hierarchical models with neural networks for structured representation but do not discuss factorized priors but rather other types of priors such as dynamical systems.”
• Multiresolution Recurrent Neural Networks: An Application to Dialogue Response Generation (2016, background): “2015), the Variational Recurrent Autoencoder (Fabius and van Amersfoort 2015) and the Structured Variational Autoencoder (SVAE) (Johnson et al. 2016).”
• Conditional Variational Autoencoder for Prediction and Feature Recovery Applied to Intrusion Detection in IoT (2017, methodology): “Considering future work, after corroborating the good performance of the conditional VAE model, we plan to investigate alternative variants as ladder VAE [28] and structured VAE [29], to explore their ability to learn the probability distribution of NIDS features.”