QML Gains Tighter Generalisation Bounds via PAC-Bayesian Framework

Researchers from UPV/EHU, Warwick & FU Berlin derived the first data-dependent PAC-Bayesian generalisation bounds for dissipative layered quantum circuits, using a hybrid L1/L2 norm to better assess QML model performance beyond worst-case estimates.

#QuantumMachineLearning #PACBayesian #News

New Tight PAC-Bayesian Certificates Boost Contrastive Learning Theory

Researchers introduced tighter PAC‑Bayesian risk certificates for contrastive learning; CIFAR‑10 tests show the bounds closely match observed empirical errors. Read more: getnews.me/new-tight-pac-bayesian-c... #pacbayesian #cifar10