Learning Regular Expressions Proven Hard in PAC Model

A new Oct 2025 study proves learning regular expressions is hard in the PAC model and with membership queries, even under a uniform Boolean hypercube distribution. getnews.me/learning-regular-express... #regularexpressions #paclearning

Higher-Arity PAC Learning and New VC Dimension Theory Unveiled

A new preprint (arXiv:2010.00726) extends VC dimension to VC_n and shows bounded VC_n characterizes PAC_n learning for product measures. Published Oct 2025. Read more: getnews.me/higher-arity-pac-learnin... #vcdimension #paclearning #higherarity

PAC Learning Boosts Personalized Prediction via Sparse Linear Models

Researchers propose a method that builds a classifier per query using reference classes, achieving an excess error bound of O(opt¹/⁴) and matching accuracy on benchmarks. getnews.me/pac-learning-boosts-pers... #paclearning #personalizedprediction