1985–, Computer scientist
Also known as: Durk Kingma
Diederik P. Kingma is a Dutch computer scientist who, with Max Welling, co-invented the variational autoencoder (VAE) in 2013, providing a tractable framework for training deep generative models with continuous latent variables. With Jimmy Ba he co-developed Adam (2014), the adaptive-learning-rate optimiser that has become the default optimiser of deep learning.
Kingma's PhD thesis (Amsterdam, 2017) on variational inference and deep latent-variable models is one of the most-cited theses in modern machine learning. He has been at OpenAI, Google Brain and most recently Anthropic, contributing to language modelling, generative modelling and the broader theory of deep learning.
Related people: Max Welling, Yoshua Bengio
Works cited in this book:
- Auto-Encoding Variational Bayes (2013) (with Max Welling)
- Adam: A Method for Stochastic Optimization (2014) (with Jimmy Ba)
- Score-Based Generative Modeling through Stochastic Differential Equations (2020) (with Yang Song, Jascha Sohl-Dickstein, Abhishek Kumar, Stefano Ermon, Ben Poole)
Discussed in:
- Chapter 14: Generative Models, Generative Models