Large-Scale Cox Process Inference using Variational Fourier Features

Gaussian process modulated Poisson processes provide a flexible framework for modelling spatiotemporal point patterns. So far this had been restricted to one dimension, binning to a pre-determined grid, or small data sets of up to a few thousand data points. Here we introduce Cox process inference based on Fourier features. This sparse representation induces global rather than local constraints on the function space and is computationally efficient. This allows us to formulate a grid-free approximation that scales well with the number of data points and the size of the domain. We demonstrate that this allows MCMC approximations to the non-Gaussian posterior. We also find that, in practice, Fourier features have more consistent optimization behavior than previous approaches. Our approximate Bayesian method can fit over 100,000 events with complex spatiotemporal patterns in three dimensions on a single GPU.

[1]  Aki Vehtari,et al.  Sparse Log Gaussian Processes via MCMC for Spatial Epidemiology , 2007, Gaussian Processes in Practice.

[2]  João Gama,et al.  Predicting Taxi–Passenger Demand Using Streaming Data , 2013, IEEE Transactions on Intelligent Transportation Systems.

[3]  Arno Solin,et al.  Variational Fourier Features for Gaussian Processes , 2016, J. Mach. Learn. Res..

[4]  Neil D. Lawrence,et al.  Bayesian Gaussian Process Latent Variable Model , 2010, AISTATS.

[5]  Ryan P. Adams,et al.  Tractable nonparametric Bayesian inference in Poisson processes with Gaussian process intensities , 2009, ICML '09.

[6]  Peter J. Diggle,et al.  Spatial and spatio-temporal Log-Gaussian Cox processes:extending the geostatistical paradigm , 2013, 1312.6536.

[7]  Benjamin Recht,et al.  Random Features for Large-Scale Kernel Machines , 2007, NIPS.

[8]  Stephen J. Roberts,et al.  Efficient Bayesian Nonparametric Modelling of Structured Point Processes , 2014, UAI.

[9]  Carl E. Rasmussen,et al.  Sparse Spectrum Gaussian Process Regression , 2010, J. Mach. Learn. Res..

[10]  Adrian N. Bishop,et al.  Fast Bayesian Intensity Estimation for the Permanental Process , 2017, ICML.

[11]  Yee Whye Teh,et al.  Poisson intensity estimation with reproducing kernels , 2016, AISTATS.

[12]  James Hensman,et al.  MCMC for Variationally Sparse Gaussian Processes , 2015, NIPS.

[13]  Stephen J. Roberts,et al.  Variational Inference for Gaussian Process Modulated Poisson Processes , 2014, ICML.

[14]  Peter J. Diggle,et al.  Bayesian Inference and Data Augmentation Schemes for Spatial, Spatiotemporal and Multivariate Log-Gaussian Cox Processes in R , 2015 .

[15]  J. Møller,et al.  Log Gaussian Cox Processes , 1998 .

[16]  Alexander G. de G. Matthews,et al.  Scalable Gaussian process inference using variational methods , 2017 .

[17]  Iain Murray.  Advances in Markov chain Monte Carlo methods , 2007 .

[18]  P. Diggle.  A Kernel Method for Smoothing Point Process Data , 1985 .

[19]  James Hensman,et al.  On Sparse Variational Methods and the Kullback-Leibler Divergence between Stochastic Processes , 2015, AISTATS.

[20]  S. Flaxman,et al.  Scalable high-resolution forecasting of sparse spatiotemporal events with kernel methods: A winning solution to the NIJ “Real-Time Crime Forecasting Challenge” , 2018, The Annals of Applied Statistics.