A1039
Title: Convergence analysis for functional online learning algorithms
Authors: Xin Guo - The University of Queensland (Australia)
Lei Shi - Fudan University (China)
Zheng-Chu Guo - Zhejiang University (China) [presenting]
Abstract: Convergence analysis of online stochastic gradient descent algorithms for functional linear models is reported. Adopting the characterizations of the slope function regularity, the kernel space capacity, and the sampling process covariance operator capacity, significant improvement in the convergence rates is achieved. Both prediction and estimation problems are studied, showing that capacity assumption can alleviate the convergence rate saturation as the target function's regularity increases. It is shown that with a properly selected kernel, capacity assumptions can fully compensate for the regularity assumptions for prediction problems (but not estimation problems). This demonstrates the significant difference between functional data analysis prediction and estimation problems.
