BriefGPT.xyz
May, 2020
高效学习带有噪声的对抗稳健超平面
Efficiently Learning Adversarially Robust Halfspaces with Noise
HTML
PDF
Omar Montasser, Surbhi Goel, Ilias Diakonikolas, Nathan Srebro
TL;DR
研究在独立于分布的情况下学习对抗性鲁棒的半空间问题，在可实现的情况下，给出了对抗性扰动集的必要和充分条件，即半空间可以高效地实现鲁棒性学习，并在存在随机标签噪声时提出了一个简单的、计算效率高的算法。
Abstract
We study the problem of learning adversarially robust
halfspaces
in the
distribution-independent
setting. In the realizable setting, we provide necessary and sufficient conditions on the adversarial perturbation
→