AI RESEARCH
Computationally Efficient Replicable Learning of Parities and Applications
arXiv CS.LG
•
ArXi:2602.09499v2 Announce Type: replace We study the computational relationship between replicability (Impagliazzo [STOC `22], Ghazi [NeurIPS `21]) and other stability notions. Specifically, we focus on replicable PAC learning and its connections to differential privacy (Dwork [TCC 2006]) and to the statistical query (SQ) model (Kearns [JACM `98]). Statistically, it was known that differentially private learning and replicable learning are equivalent and strictly powerful than SQ-learning.