B1509
Title: A study of nonconvex risk minimization in statistical learning
Authors: Yunlong Feng - The State University of New York at Albany (United States) [presenting]
Abstract: Nonconvex loss functions have been more and more frequently used owing to their robustness to outliers and heavy-tailed noise. However, understanding of nonconvex loss functions, especially from a theoretical viewpoint, is still limited. Some recent efforts are reported by focusing on bounded nonconvex losses. First, it is shown that in the context of empirical risk minimization, bounded nonconvex loss functions can be interpreted from a minimum distance estimation viewpoint. Second, results on the prediction ability of estimators resulting from bounded nonconvex losses are also provided and discussed.