Non-Stationary Spectral Kernels
arXiv:1705.08736
Abstract
We propose non-stationary spectral kernels for Gaussian process regression. We propose to model the spectral density of a non-stationary kernel function as a mixture of input-dependent Gaussian process frequency density surfaces. We solve the generalised Fourier transform with such a model, and present a family of non-stationary and non-monotonic kernels that can learn input-dependent and potentially long-range, non-monotonic covariances between inputs. We derive efficient inference using model whitening and marginalized posterior, and show with case studies that these kernels are necessary when modelling even rather simple time series, image or geospatial data with non-stationary characteristics.
16 pages, 5 figures
Cited by in corpus (20)
- Uncertainty Quantification in Machine Learning for Engineering Design and Health Prognostics: A Tutorial
- Differentiable Compositional Kernel Learning for Gaussian Processes
- Random Features for Kernel Approximation: A Survey on Algorithms, Theory, and Beyond
- Automated Spectral Kernel Learning
- Large-scale Heteroscedastic Regression via Gaussian Process
- Marginalised Gaussian Processes with Nested Sampling
- Sparse Spectrum Warped Input Measures for Nonstationary Kernel Learning
- Recent Advances in Data-Driven Wireless Communication Using Gaussian Processes: A Comprehensive Survey
- Marginalising over Stationary Kernels with Bayesian Quadrature
- Convolutional Spectral Kernel Learning
- Approximate Inference for Spectral Mixture Kernel
- A Temporal Kernel Approach for Deep Learning with Continuous-time Information
- Adaptive Inducing Points Selection For Gaussian Processes
- State-space deep Gaussian processes with applications
- End-to-End Probabilistic Inference for Nonstationary Audio Analysis
- A Nonstationary Designer Space-Time Kernel
- The Minecraft Kernel: Modelling correlated Gaussian Processes in the Fourier domain
- Scalable Hybrid HMM with Gaussian Process Emission for Sequential Time-series Data Clustering
- Time-Variant Variational Transfer for Value Functions
- Gaussian Processes with Skewed Laplace Spectral Mixture Kernels for Long-term Forecasting