Open AccessProceedings Article
Deep Neural Networks as Gaussian Processes
Jaehoon Lee,Yasaman Bahri,Roman Novak,Samuel S. Schoenholz,Jeffrey Pennington,Jascha Sohl-Dickstein +5 more
TLDR
The exact equivalence between infinitely wide deep networks and GPs is derived and it is found that test performance increases as finite-width trained networks are made wider and more similar to a GP, and thus that GP predictions typically outperform those of finite- width networks.Abstract:
It has long been known that a single-layer fully-connected neural network with an i.i.d. prior over its parameters is equivalent to a Gaussian process (GP), in the limit of infinite network width. This correspondence enables exact Bayesian inference for infinite width neural networks on regression tasks by means of evaluating the corresponding GP. Recently, kernel functions which mimic multi-layer random neural networks have been developed, but only outside of a Bayesian framework. As such, previous work has not identified that these kernels can be used as covariance functions for GPs and allow fully Bayesian prediction with a deep neural network.
In this work, we derive the exact equivalence between infinitely wide deep networks and GPs. We further develop a computationally efficient pipeline to compute the covariance function for these GPs. We then use the resulting GPs to perform Bayesian inference for wide deep neural networks on MNIST and CIFAR-10. We observe that trained neural network accuracy approaches that of the corresponding GP with increasing layer width, and that the GP uncertainty is strongly correlated with trained network prediction error. We further find that test performance increases as finite-width trained networks are made wider and more similar to a GP, and thus that GP predictions typically outperform those of finite-width networks. Finally we connect the performance of these GPs to the recent theory of signal propagation in random neural networks.read more
Citations
More filters
Posted Content
Uncertainty in Neural Networks: Bayesian Ensembling.
TL;DR: This work proposes one modification to the usual ensembling process that does result in Bayesian behaviour: regularising parameters about values drawn from a prior distribution.
Proceedings Article
Infinite attention: NNGP and NTK for deep attention networks
TL;DR: A rigorous extension of results to NNs involving attention layers is provided, showing that unlike single- head attention, which induces non-Gaussian behaviour, multi-head attention architectures behave as GPs as the number of heads tends to infinity.
Proceedings Article
Deep learning generalizes because the parameter-function map is biased towards simple functions
TL;DR: The PAC-Bayes theorem can be used to guarantee good expected generalization for target functions producing high-likelihood training sets as mentioned in this paper, which helps explain why deep neural networks generalize well on real world problems.
Posted Content
Random Features for Kernel Approximation: A Survey on Algorithms, Theory, and Beyond
TL;DR: This survey systematically review the work on random features from the past ten years, and discusses the relationship between random features and modern over-parameterized deep neural networks, including the use of random features in the analysis DNNs as well as the gaps between current theoretical and empirical results.
Posted Content
Dataset Meta-Learning from Kernel Ridge-Regression
TL;DR: This work introduces the novel concept of $\epsilon$-approximation of datasets, obtaining datasets which are much smaller than or are significant corruptions of the original training data while maintaining similar model performance.
References
More filters
Proceedings Article
Adam: A Method for Stochastic Optimization
Diederik P. Kingma,Jimmy Ba +1 more
TL;DR: This work introduces Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments, and provides a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework.
Book
Bayesian learning for neural networks
TL;DR: Bayesian Learning for Neural Networks shows that Bayesian methods allow complex neural network models to be used without fear of the "overfitting" that can occur with traditional neural network learning methods.
Journal ArticleDOI
A Unifying View of Sparse Approximate Gaussian Process Regression
TL;DR: A new unifying view, including all existing proper probabilistic sparse approximations for Gaussian process regression, relies on expressing the effective prior which the methods are using, and highlights the relationship between existing methods.
Journal Article
In Defense of One-Vs-All Classification
Ryan Rifkin,Aldebaro Klautau +1 more
TL;DR: It is argued that a simple "one-vs-all" scheme is as accurate as any other approach, assuming that the underlying binary classifiers are well-tuned regularized classifiers such as support vector machines.
Proceedings Article
Gaussian processes for Big data
TL;DR: In this article, the authors introduce stochastic variational inference for Gaussian process models, which enables the application of Gaussian Process (GP) models to data sets containing millions of data points.