Kernel methods in PyTorch

Fast implementations of standard utilities

Installation

pip install -I git+https://github.com/parthe/torchkernels

Requires a PyTorch installation

Stable behavior

Currently this code has been tested with n=10,000 samples.
with Python 3.9 and PyTorch >= 1.13

Test installation with Laplacian kernel

import torch
from torchkernels.kernels.radial import laplacian, LaplacianKernel

n = 300 # number of samples
p = 200 # number of centers
d = 100  # dimensions

is_cuda = torch.cuda.is_available()
DEV = torch.device("cuda") if is_cuda else torch.device("cpu")    

X = torch.randn(n, d, device=DEV)
Z = torch.randn(p, d, device=DEV)

kernel_matrix1 = laplacian(X, Z, bandwidth=1.)

K = LaplacianKernel(bandwidth=1.)
kernel_matrix2 = K(X, Z)

torch.testing.assert_close(kernel_matrix1, kernel_matrix2, msg='Laplacian test failed')
print('Laplacian test complete!')

Currently supported Kernels

• Laplacian, Gaussian, Dispersal (Exponential power kernel)
• Normalized dot-product kernel for arbitrary functions
• Neural Network Gaussian Process (NNGP) and Tangent Kernel (NTK) with ReLU activations

Other utilities

• top eigenvectors of kernel matrix
• Random feature maps for the Laplacian, Gaussian, Matern and the ExpPower kernel