Error analysis of distributed algorithm for large scale data classification.

The distributed algorithm is an important and basic approach, and it is usually used for large scale data processing. This paper aims to error analysis of distributed algorithm for large scale data classification generated from Tikhonov regularization schemes associated with varying Gaussian kernels and convex loss functions. The main goal is to provide fast convergence rates for the excess misclassification error. The number of subsets randomly divided from a large scale datasets is determined to guarantee that the distributed algorithm have lower time complexity and memory complexity. [ABSTRACT FROM AUTHOR]