{"id":"arxiv/2601.13436","url":"https://arxiv.org/abs/2601.13436","tags":["stat.ML","cs.LG","eess.SP","eess.SY","math.ST"],"score":0,"title":"Distribution-Free Confidence Ellipsoids for Ridge Regression with PAC Bounds","domain":"stat.ML","source":"arxiv","authors":["Szabolcs Szentpéteri","Balázs Csanád Csáji"],"reviews":0,"subject":"stat.ML","summary":"Linearly parametrized models are widely used in control and signal processing, with the least-squares (LS) estimate being the archetypical solution. When the input is insufficiently exciting, the LS problem may be unsolvable or numerically unstable. This issue can be resolved through regularization, typically with ridge regression. Although regularized estimators reduce the variance error, it remains important to quantify their estimation uncertainty. A possible approach for linear regression is to construct confidence ellipsoids with the Sign-Perturbed Sums (SPS) ellipsoidal outer…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-01-19","reactions":{}}