2009/12/31 by Shahar Mendelson, Joseph Neeman
Computer Science · Mathematics · #Neural Networks and Applications #Numerical methods in inverse problems #Stochastic Gradient Optimization Techniques #math.ST #msc:60G99 #msc:68Q32 #stat.TH
paper · pdf · doi:10.1214/09-aos728
published as Annals of Statistics 2010, Vol. 38, No. 1, 526-565 · Published in at http://dx.doi.org/10.1214/09-AOS728 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)
openalex publication_date 2009/12/31 · arxiv created 2010/01/13 · arxiv updated 2010/01/14 · openalex created_date 2016/06/24 · openalex updated_date 2026/08/01
Under mild assumptions on the kernel, we obtain the best known error rates in a regularized learning scenario taking place in the corresponding reproducing kernel Hilbert space (RKHS). The main novelty in the analysis is a proof that one can use a regularization term that grows significantly slower than the standard quadratic growth in the RKHS norm.