Faster Kernel Matrix Algebra via Density Estimation
Arturs Backurs * 1 Piotr Indyk * 2 Cameron Musco * 3 Tal Wagner * 4
Abstract et al. (2008) for an overview.
We study fast algorithms for computing funda- Kernel methods typically operate using a kernel matrix.
mental properties of a positive semidefinite ker- Given n vectors x1 , . . . , xn ∈ Rd , the kernel matrix K ∈
nel matrix K ∈ Rn×n corresponding to n points ...
附件列表