1 paper
Sergei Tikhonov, Arsen Vasilyan
We study the problem of computationally efficient proper agnostic learning of multidimensional concept classes under the Gaussian distribution. In this setting, given i.i.d. labele…