For interpolating kernel machines, the minimum norm ERM solution is the most stable

TitleFor interpolating kernel machines, the minimum norm ERM solution is the most stable
Publication TypeCBMM Memos
Year of Publication2020
AuthorsRangamani, A, Rosasco, L, Poggio, T
Number108
Date Published06/2020
Abstract

We study the average CVloo stability of kernel ridge-less regression and derive corresponding risk bounds. We show that the interpolating solution with minimum norm has the best CVloo stability, which in turn is controlled by the condition number of the empirical kernel matrix. The latter can be characterized in the asymptotic regime where both the dimension and cardinality of the data go to infinity. Under the assumption of random kernel matrices, the corresponding test error follows a double descent curve.

arXiv

https://arxiv.org/abs/2006.15522

DSpace@MIT

https://hdl.handle.net/1721.1/125927

CBMM Memo No:  108

Associated Module: 

CBMM Relationship: 

  • CBMM Funded