Tutorial on Variational Autoencoders

Carl Doersch
6/19/2016
stat.MLcs.LG

Official code: cdoersch/vae_tutorial (523 stars, MIT, last commit 2016-12-22).

Abstract

In just three years, Variational Autoencoders (VAEs) have emerged as one of the most popular approaches to unsupervised learning of complicated distributions. VAEs are appealing because they are built on top of standard function approximators (neural networks), and can be trained with stochastic gradient descent. VAEs have already shown promise in generating many kinds of complicated data, including handwritten digits, faces, house numbers, CIFAR images, physical models of scenes, segmentation, and predicting the future from static images. This tutorial introduces the intuitions behind VAEs, explains the mathematics behind them, and describes some empirical behavior. No prior knowledge of variational Bayesian methods is assumed.

View on arXivView PDF

Code Implementations(1)

cdoersch/vae_tutorialOfficial100%
MIT

Caffe code to accompany my Tutorial on Variational Autoencoders

523135PythonJun 19, 20169 years agoMIT

Similar Papers

Learning Distributions via Monte-Carlo Marginalization

83% match2023

Deep Unsupervised Learning using Nonequilibrium Thermodynamics

82% match1 repo5462015

Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data

82% match1 repo142016

On GANs and GMMs

81% match1 repo672018

Is Deep Learning Network Necessary for Image Generation?

81% match2023

Introduction to Stochastic Differential Equations for Generative Machine Learning: A Variational Perspective

81% match2026

Cite this paper

@article{doersch2016tutorial,
  title  = {Tutorial on Variational Autoencoders},
  author = {Carl Doersch},
  year   = {2016},
  eprint = {1606.05908},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1606.05908}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1606.05908.svg)](https://codeofpaper.com/paper/1606.05908)

Discussion