University of Cambridge > > Machine Learning Reading Group @ CUED > Recent advances in the theory and applications of VAEs

Recent advances in the theory and applications of VAEs

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact jg801.

Since their development by Kingma[1] the VAE has proven to be a flexible and powerful framework for latent variable modelling. Sitting at the crossroads of classical statistical learning and deep learning, it has made an impact in many problems, from classical tasks in computer vision such as image generation, compression, super-resolution, inpainting, 3D object synthesis to other tasks such as semi-supervised learning, natural language modelling, sentence interpolation, and even practical applications to medical imaging.

We will start with a minimal introduction to latent variable models and the structure and definition of a VAE , before presenting some examples. We will then discuss some of the topics of interest in the research surrounding VAEs, looking to various techniques meant to reduce the inference gap, as well as studying the usefulness of the VAE for learning latent representations for downstream tasks. If time allows, we will present some more off-beat results.

Auto-encoding variational Bayes; D. Kingma, M. Welling:

This talk is part of the Machine Learning Reading Group @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity