ReLU Regression with Massart Noise
This addresses a fundamental challenge in robust learning for neural networks, offering a solution for scenarios with semi-random label noise, though it is incremental as it builds on known noise models.
The paper tackles the problem of ReLU regression under Massart noise, where labels are corrupted with bounded probability, and presents an efficient algorithm that achieves exact parameter recovery under mild anti-concentration assumptions, outperforming naive regression methods on synthetic and real data.
We study the fundamental problem of ReLU regression, where the goal is to fit Rectified Linear Units (ReLUs) to data. This supervised learning task is efficiently solvable in the realizable setting, but is known to be computationally hard with adversarial label noise. In this work, we focus on ReLU regression in the Massart noise model, a natural and well-studied semi-random noise model. In this model, the label of every point is generated according to a function in the class, but an adversary is allowed to change this value arbitrarily with some probability, which is {\em at most} $η< 1/2$. We develop an efficient algorithm that achieves exact parameter recovery in this model under mild anti-concentration assumptions on the underlying distribution. Such assumptions are necessary for exact recovery to be information-theoretically possible. We demonstrate that our algorithm significantly outperforms naive applications of $\ell_1$ and $\ell_2$ regression on both synthetic and real data.