Locally Regularized Orthogonal Least Squares and D-Optimality Experimental Design for RBF regression modeling
統計科学セミナー
開催期間
2010.10.29(金)
16:00 ~ 17:00
16:00 ~ 17:00
場所
伊都キャンパス 伊都図書館3階 中セミナー室4
講演者
Jan Dolinsky (九州大学 大学院数理学研究院)
概要
Locally Regularized Orthogonal Least Squares (LROLS) combined
with D-Optimality Experimental Design is an efficient identification
algorithm. This algorithm is used for RBF regression modeling with
high-dimensional regressor. Numerical experiments show that the
algorithm performs well and converges fast in various situations.