An Introduction to Variational Autoencoders
By Diederik P. Kingma, Max Welling. 2019
3,108 citations recorded by the source corpus.
Variational autoencoders provide a principled framework for learning deep latent-variable models and corresponding inference models. In this work, we provide an introduction to variational autoencoders and some important extensions.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Statistical and Nonlinear Physics, 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 Artificial Intelligence
- Medhi Bennis
- Wojciech Samek
- Jihong Park
- K. Müller
- Grgoire Montavon
- Yoshua Bengio
- Ming-Hsuan Yang
- Merouane Debbah
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Xiaogang Wang
- Trevor Darrell
- Bernt Schiele
- Andrea Vedaldi
- Jitendra Malik
In Statistical and Nonlinear Physics
- Wojciech Samek
- Eldad Haber
- Lars Ruthotto
- Grgoire Montavon
- Ming-Hsuan Yang
- Bin Cui
- Yue Zhao
- Wentao Zhang
In Generative Models
In Probabilistic & Bayesian Methods
Citations in the corpus
- Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images (2020, background): “See Kingma & Welling (2019) for an in-depth introduction.”
- Prototype Augmentation and Self-Supervision for Incremental Learning (2021, neutral): “Nevertheless, it is inefficient to train big generative models such as GAN [17, 3] and autoencoder [27, 25] for complex datasets (e.”
- Autoregressive Denoising Diffusion Models for Multivariate Probabilistic Time Series Forecasting (2021, positive): “Unlike in variational autoen-coders (Kingma & Welling, 2019) the approximate posterior q ( x 1: N | x 0 ) , is not trainable but fixed to a Markov chain (called the forward process) that gradually adds Gaussian noise to the signal: ) .”
- A Comprehensive Overview and Comparative Analysis on Deep Learning Models: CNN, RNN, LSTM, GRU (2023, background): “In terms of Bayes' rule, the recognition model is roughly the inverse of the generative model [63].”
- RODIN: A Generative Model for Sculpting 3D Digital Avatars Using Diffusion (2022, methodology): “Early 3D generation works rely on either GAN [17] or VAE [29] to model the distribution of 3D shape representation like voxel grids [6, 70], point clouds [1, 7, 31, 72], mesh [33, 63] and implicit neural representation [42, 60].”
- EDGE: Editable Dance Generation From Music (2022, positive): “Much like previous generative approaches like VAE [28] and GAN [12], diffusion models are also capable of conditional generation.”
- Efficient-VDVAE: Less is more (2022, neutral): “For a derivation of the ELBO and a more in depth introduction to VAEs, see [17].”
- Deep Generative Models in Engineering Design: A Review (2021, background): “are encouraged to refer to the literature [5] for a more detailed overview.”
- An introduction to deep generative modeling (2021, methodology): “In Section 4, we introduce variational autoencoders [27, 40, 28].”
- Informed Machine Learning – A Taxonomy and Survey of Integrating Prior Knowledge into Learning Systems (2019, methodology): “Directions responding to this are variational methods with origins in optimization theory and functional analysis [146] and variational neural networks [147].”
- Deep Learning: A Comprehensive Overview on Techniques, Taxonomy, Applications and Research Directions (2021, background): “Regularized autoencoders such as sparse, denoising, and contractive are useful for learning representations for later classification tasks [119], while variational autoencoders can be used as generative models [56], discussed below.”
- Continual Learning Of Predictive Models In Video Sequences Via Variational Autoencoders (2020, methodology): “To optimize them, the VAE maximizes the sum of the lower bound on the marginal likelihood of each observation x belonging to a given the dataset D [17, 18].”
- LIFT: Language-Interfaced Fine-Tuning for Non-Language Machine Learning Tasks (2022, methodology): “Generative models have been widely used in computer vision [56, 57, 58].”
- Diffusion Models for Medical Anomaly Detection (2022, background): “Related Work In classical anomaly detection, autoencoders [28,13] are trained on data of healthy subjects.”
- Unleashing the Power of Edge-Cloud Generative AI in Mobile Networks: A Survey of AIGC Services (2023, neutral): “3) Variational Autoencoder: The VAE [73] is a type of generative models that consist of two primary components: an encoder and a decoder network.”
- Can deep learning beat numerical weather prediction? (2021, background): “For further information on VAE, we refer to [47,55].”