The paper shows that for learning algorithms based on preferential compression, the risk can be bounded by the compression complexity term λk/N plus an O(1/√N) term, eliminating the previously assumed logarithmic factor. This bound holds uniformly over all possible values of λk and matches the law‑of‑large‑numbers rate of empirical risk estimation. The authors also demonstrate that the leading λk/N term is necessary and analyze when empirical risk converges at O(1/√N) versus slower or inconsistent rates, while the complexity‑based estimator retains the O(1/√N) rate.
By Simone Garatti, Marco C. Campi
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: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
arXiv:2609.24260v1 Announce Type: cross
Abstract: We study the problem of \emph{adversarially robust} PAC learning. In this framework, the learner observes independent samples from an unknown distrib...
By Steve Hanneke, Amirreza Shaeiri
arXiv:2608. 06337v1 Announce Type: cross Abstract: A monotone adversary observes an i.
By Anay Mehrotra
arXiv:2609.36030v1 Announce Type: new
Abstract: The lack of rigorous safety and performance certificates remains a key bottleneck to the deployment of modern learning-based methods. Sample compressio...
By Dario Paccagnan, Marius Tirlea
We study learning a discrete distribution under person-level differential privacy in the batched setting. Given $n$ users, each contributing $m$ i.
arXiv:2606. 29331v1 Announce Type: new Abstract: Scientific discovery via symbolic regression is often viewed as statistically and computationally intractable because the hypothesis space of expressions grows combinatorially with depth.
By \c{S}uayp Talha Kocabay, Talha R\"uzgar Akku\c{s}, Kerem Yal\c{c}{\i}n
arXiv:2608. 10869v1 Announce Type: new Abstract: Worst-case multiclass bounds do not become smaller when the best classifier is already nearly correct: what is missing is an optimistic rate, a guarantee whose fluctuation scales with the oracle risk itself.
By Xiaoyu Li, Andi Han, Jiaojiao Jiang, Junbin Gao
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:2605.13684v2 Announce Type: replace
Abstract: We study the optimal scale at which real-valued function classes exhibit uniform convergence and learnability. Our main result establishes a scale-...
By Shashaank Aiyer, Yishay Mansour, Shay Moran, Han Shao, Tom Waknine
arXiv:2601. 01496v3 Announce Type: replace-cross Abstract: In this paper, we settle the problem of learning optimal linear contracts from data in the offline setting, where agent types are drawn from an unknown distribution and the principal's goal is to design a contract that maximizes her expected utility.
By Mikael M{\o}ller H{\o}gsgaard