WD Wei Dai Statistics · ML · Scientific Computing

Interactive lab

Kernel Playground

Compare how different kernels change covariance structure and prior sample paths before fitting any data.

This explainer focuses on the prior side of Gaussian-process modeling. Before any observations arrive, the kernel already encodes a strong set of assumptions about smoothness, periodicity, and long-range dependence.

Use the controls to compare how different kernels reshape both the covariance surface and the kinds of functions the prior is willing to generate.

Concept: Prior geometry, inductive bias, and sample-path behavior across RBF, periodic, linear, and Matern kernels.
1.10
0.90
1.60
0.00

Prior sample paths

Each line is a fresh draw from the prior induced by the selected kernel.

Sample paths from the selected Gaussian-process prior

Covariance matrix

The heatmap shows how strongly each pair of input locations co-varies under the kernel.

Covariance matrix heatmap for the selected kernel

Open to new collaborations

Looking for work that connects statistical rigor with practical systems.

View CV Email