TL;DR本文基于 Sum of Squares 方法,探讨了用于高维下学习高度分离的高斯混合物和鲁棒均值估计的新有效算法,进一步优化了以往算法的统计保障。通过在高度分离的高斯混合物和穿插噪音后的子高斯分布上实现均值估计,我们的方法多次突破优化算法的极限。
Abstract
We use the sum of squares method to develop new efficient algorithms for
learning well-separated mixtures of Gaussians and robust mean estimation, both
in high dimensions, that substantially improve upon the stat