Online Learning as Stochastic Approximation of Regularization Paths: Optimality and Almost-sure Convergence
Yao, Y.; Tarres, Pierre (2014), Online Learning as Stochastic Approximation of Regularization Paths: Optimality and Almost-sure Convergence, IEEE Transactions on Information Theory, PP, 99. http://dx.doi.org/10.1109/TIT.2014.2332531
TypeArticle accepté pour publication ou publié
Journal nameIEEE Transactions on Information Theory
MetadataShow full item record
Abstract (EN)In this paper, an online learning algorithm is proposed as sequential stochastic approximation of a regularization path converging to the regression function in reproducing kernel Hilbert spaces (RKHSs). We show that it is possible to produce the best known strong (RKHS norm) convergence rate of batch learning, through a careful choice of the gain or step size sequences, depending on regularity assumptions on the regression function. The corresponding weak (mean square distance) convergence rate is optimal in the sense that it reaches the minimax and individual lower rates in the literature. In both cases we deduce almost sure convergence, using Bernstein-type inequalities for martingales in Hilbert spaces. To achieve this we develop a bias-variance decomposition similar to the batch learning setting; the bias consists in the approximation and drift errors along the regularization path, which display the same rates of convergence, and the variance arises from the sample error analysed as a (reverse) martingale difference sequence. The rates above are obtained by an optimal trade-off between the bias and the variance.
Subjects / KeywordsUpper bound; Probabilistic logic; Kernel; Hilbert space; Educational institutions; Convergence; Approximation methods
Showing items related by title and author.
Error estimates and convergence rates for the stochastic homogenization of Hamilton-Jacobi equations Armstrong, Scott N.; Cardaliaguet, Pierre; Souganidis, Panagiotis E. (2014) Article accepté pour publication ou publié
Homogenization of degenerate second-order PDE in periodic and almost periodic environments and applications Lions, Pierre-Louis; Souganidis, Panagiotis E. (2005) Article accepté pour publication ou publié