A sub-constant improvement in approximating the positive semidefinite Grothendieck problem.

Roy Frostig Sida I. Wang

Published in: CoRR (2014)

Keyphrases

positive semidefinite
semidefinite programming
kernel matrix
similarity function
kernel methods
kernel function
metric learning
semidefinite
linear programming
polynomial kernels
machine learning
linear combination
feature selection
primal dual
positive semidefinite matrices