Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
7548738 | Statistics & Probability Letters | 2018 | 7 Pages |
Abstract
This paper considers a coefficient-based additive model with the âq-regularizer (1â¤qâ¤2). Error bounds are established for the proposed model by integrating the stepping stone technique and the concentration estimate with empirical covering numbers. From error analysis, we obtain a sharp learning rate that can be arbitrarily close to O(nϵâ1) under mild conditions.
Keywords
Related Topics
Physical Sciences and Engineering
Mathematics
Statistics and Probability
Authors
Yanfang Tao, Biqin Song, Luoqing Li,