Jump to Main Content
Regression with stagewise minimization on risk function
- Yoshida, Takuma, Naito, Kanta
- Computational statistics & data analysis 2019 v.134 pp. 123-143
- algorithms, risk
- This paper studies a curve estimation based on empirical risk minimization. The estimator is composed as a convex combination of words (learners) in a dictionary. A word is selected in each step of the proposed stagewise algorithm, which minimizes a certain divergence measure. A non-asymptotic error bound of the estimator is developed, and it is shown that the error bound becomes sharp as the number of iterations of the algorithm increases. A simulation study and real data example confirm the performance of the estimator.