Omnipredicting Single-Index Models with Multi-Index Models

L Hu, K Tian, C Yang - arxiv preprint arxiv:2411.13083, 2024 - arxiv.org
Recent work on supervised learning [GKR+ 22] defined the notion of omnipredictors, ie,
predictor functions $ p $ over features that are simultaneously competitive for minimizing a …

Sample and Computationally Efficient Robust Learning of Gaussian Single-Index Models

P Wang, N Zarifis, I Diakonikolas… - arxiv preprint arxiv …, 2024 - arxiv.org
A single-index model (SIM) is a function of the form $\sigma (\mathbf {w}^{\ast}\cdot\mathbf
{x}) $, where $\sigma:\mathbb {R}\to\mathbb {R} $ is a known link function and $\mathbf …

Learning noisy halfspaces with a margin: Massart is no harder than random

G Chandrasekaran, V Kontonis, K Stavropoulos… - arxiv preprint arxiv …, 2025 - arxiv.org
We study the problem of PAC learning $\gamma $-margin halfspaces with Massart noise.
We propose a simple proper learning algorithm, the Perspectron, that has sample …

Learning a Single Neuron Robustly to Distributional Shifts and Adversarial Label Noise

S Li, S Karmalkar, I Diakonikolas… - arxiv preprint arxiv …, 2024 - arxiv.org
We study the problem of learning a single neuron with respect to the $ L_2^ 2$-loss in the
presence of adversarial distribution shifts, where the labels can be arbitrary, and the goal is …

Robustly Learning Monotone Generalized Linear Models via Data Augmentation

N Zarifis, P Wang, I Diakonikolas… - arxiv preprint arxiv …, 2025 - arxiv.org
We study the task of learning Generalized Linear models (GLMs) in the agnostic model
under the Gaussian distribution. We give the first polynomial-time algorithm that achieves a …

Convergence of for Gradient-Based Algorithms in Zero-Sum Games without the Condition Number: A Smoothed Analysis

I Anagnostides, T Sandholm - arxiv preprint arxiv:2410.21636, 2024 - arxiv.org
Gradient-based algorithms have shown great promise in solving large (two-player) zero-sum
games. However, their success has been mostly confined to the low-precision regime since …