We study the problem of \emph{adversarially robust} PAC learning. In this framework, the learner observes independent samples from an unknown distribution over $\mathcal{X} \times \{0,1\}$, as in clas...
arXiv:2608. 04686v1 Announce Type: new Abstract: We study distributionally robust PAC learning for the $0$--$1$-loss, where adversarial perturbations of the data distribution are constrained by a Cressie--Read divergence of order $k>1$ and radius $\rho\geq 0$.
By Elad Aigner-Horev, Daniel Rosenberg, Roi Weiss
arXiv:2608. 06363v1 Announce Type: cross Abstract: Let $H\subseteq\{-1,+1\}^X$ be a class of finite VC dimension $d\ge1$.
By Markus Engelund Mathiasen, Jian Qian, Nikita Zhivotovskiy
arXiv:2608. 06337v1 Announce Type: cross Abstract: A monotone adversary observes an i.
By Anay Mehrotra
The paper revisits realizable multiclass PAC learning with bandit feedback, correcting a previously claimed lower bound on sample complexity. It introduces a new anchored dimension, “aBDS,” and establishes a constant‑free three‑part lower bound, while also providing tighter upper bounds that eliminate dependence on the total label count. The authors demonstrate that the optimal sample complexity can vary dramatically even among classes with identical dimensional profiles, revealing a confidence direct‑sum phenomenon and a rank‑saturation phase transition.
By Guangjian Zhang
arXiv:2611. 13210v1 Announce Type: cross Abstract: We study the problem of agnostic PAC learning under deterministic labels, where the examples are labeled by an arbitrary deterministic function, which need not belong to the concept class $C$.
By Shai Ben-David, Steve Hanneke, Farnam Mansouri, Amirreza Shaeiri
arXiv:2610. 26944v2 Announce Type: replace Abstract: Strict replicability requires two independent training samples, processed with the same random seed, to produce exactly the same hypothesis.
By Guangjian Zhang
The paper investigates realizable PAC learning under approximate differential privacy and develops a framework for establishing lower bounds on sample complexity. It demonstrates that classes sharing the same VC and Littlestone dimensions can have different private sample complexities, introducing the Independent Class Packing (ICP) parameter to capture this distinction. Using ICP, the authors derive tight lower bounds for several hypothesis classes, including axis-aligned rectangles, unions of intervals, linear subspaces, and improved bounds for halfspaces and convex polygons, and show that ICP characterizes private PAC sample complexity for product classes up to logarithmic factors.
By Prathamesh Dharangutte, Amartya Sanyal
arXiv:2608. 13514v1 Announce Type: cross Abstract: We revisit the problem of learning predictors robust to adversarial examples at test-time.
By Omar Montasser
The paper studies learning from runs that stop at their first incorrect decision, revealing only the queries and correct labels up to that point. It analyzes how many such runs are required to learn a hypothesis from a fixed binary class of VC dimension at most $d$, providing bounds on the number of runs and distinct pilots needed to achieve a risk estimate within ε of the optimal risk. The results include upper bounds on sample complexity, treatment of unknown cost ceilings, and matching lower bounds for finite classes.
By Michal \v{C}ert\'ik
The paper investigates how well practical machine‑learning algorithms approximate Solomonoff’s universal predictor when trained by minimizing log‑loss over diverse datasets. It establishes tight bounds on the worst‑case increase in prediction error caused by approximating the Bayesian mixture predictor, showing that while the excess error can be large in low‑prior environments, it remains bounded on average and under distribution shifts. The authors also extend their analysis to finite‑sample settings relevant to neural networks and Transformers.
By Qian Li, Shang-Hua Teng
We revisit the problem of learning predictors robust to adversarial examples at test-time. We prove that VC classes are adversarially robustly learnable with sample complexity linear in the VC dimension $d$, providing an exponential improvement over the previous upper bound of Montasser, Hanneke, and Srebro (2019).