Skip to content

Codes related to the experiments shown in the ICML22 paper: 'Failure and success of the spectral bias prediction for Laplace Kernel Ridge Regression: the case of low-dimensional data'

Notifications You must be signed in to change notification settings

UmbertoTomasini/KRR_FailureSuccess_Spectral_bias

Repository files navigation

'krr_synthetic_1_dim'

Training and testing kernel ridge regression on the synthetic task in 1 dimension, as described in Section 2.
The sampling of test points is not trivial and it is explained in Appendix F.1.

'krr_synthetic_high_dim'

Training and testing kernel ridge regression on the synthetic task in >1 dimension and \xi =0, as described in Section 2.
Further details in Appendix F.2.

'krr_MNIST' and 'krr_CIFAR'

Training and testing kernel ridge regression on binary MNIST and binary CIFAR10.

'kare_testErr'

KARE prediction, as explained in Appendix B.



 

About

Codes related to the experiments shown in the ICML22 paper: 'Failure and success of the spectral bias prediction for Laplace Kernel Ridge Regression: the case of low-dimensional data'

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published