D
Diederik P. Kingma
Researcher at Google
Publications - 45
Citations - 173232
Diederik P. Kingma is an academic researcher from Google. The author has contributed to research in topics: Inference & Artificial neural network. The author has an hindex of 27, co-authored 42 publications receiving 130871 citations. Previous affiliations of Diederik P. Kingma include OpenAI & University of Amsterdam.
Papers
More filters
Proceedings Article
Learning Sparse Neural Networks through L_0 Regularization
TL;DR: In this article, a collection of non-negative stochastic gates, which collectively determine which weights to set to zero, is proposed to prune the network during training by encouraging weights to become exactly zero.
Proceedings Article
PixelCNN++: Improving the PixelCNN with Discretized Logistic Mixture Likelihood and Other Modifications
TL;DR: This work discusses the implementation of PixelCNNs, a recently proposed class of powerful generative models with tractable likelihood that contains a number of modifications to the original model that both simplify its structure and improve its performance.
Posted Content
Variational Dropout and the Local Reparameterization Trick
TL;DR: In this article, a local reparameterization technique for variational Bayesian inference of a posterior over model parameters, while retaining parallelizability, has been proposed to translate uncertainty about global parameters into local noise that is independent across datapoints in the minibatch.
Proceedings Article
Variational dropout and the local reparameterization trick
TL;DR: This work proposes variational dropout, a generalization of Gaussian dropout where the dropout rates are learned, often leading to better models, and allows inference of more flexibly parameterized posteriors.
Posted Content
Learning Sparse Neural Networks through $L_0$ Regularization
TL;DR: A practical method for L_0 norm regularization for neural networks: pruning the network during training by encouraging weights to become exactly zero, which allows for straightforward and efficient learning of model structures with stochastic gradient descent and allows for conditional computation in a principled way.