Loading timeline…
20132010s
Diederik P. Kingma
Co-invented the variational autoencoder and the Adam optimizer — two tools almost every deep learning project uses.
Organizations
University of AmsterdamOpenAIGoogle Brain
Major Achievements
- •Co-authored 'Auto-Encoding Variational Bayes' (2013) with Max Welling, introducing the variational autoencoder.
- •Introduced the reparameterisation trick, which made gradient-based training of latent-variable models practical.
- •Co-created Adam (2015), the default optimizer for deep learning.
- •Developed Glow and other normalising-flow models for exact-likelihood generation.
Key Papers
- Auto-Encoding Variational Bayes
Diederik P. Kingma, Max Welling · ICLR 2014 · 2013
- Adam: A Method for Stochastic Optimization
Diederik P. Kingma, Jimmy Ba · ICLR 2015 · 2014