Online sparse Gaussian process regression using FITC and PITC approximations

被引：31
作者：
Bijl, Hildo [1 ]
van Wingerden, Jan-Willem [1 ]
Schon, Thomas B. [2 ]
Verhaegen, Michel [1 ]
机构：
[1] Delft Univ Technol, Delft Ctr Syst & Control, NL-2600 AA Delft, Netherlands
[2] Uppsala Univ, Dept Informat Technol, Uppsala, Sweden
来源：
IFAC PAPERSONLINE | 2015年 / 48卷 / 28期
关键词：
D O I：
10.1016/j.ifacol.2015.12.212
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
We provide a method which allows for online updating of sparse Gaussian Process (GP) regression algorithms for any set of inducing inputs. This method is derived both for the Fully Independent Training Conditional (FITC) and the Partially Independent Training Conditional (PITC) approximation, and it allows the inclusion of a new measurement point x(n+1) in O(m(2)) time, with m denoting the size of the set of inducing inputs. Due to the online nature of the algorithms, it is possible to forget earlier measurement data, which means that also the memory space required is O(m(2)), both for FITC and PITC. We show that this method is able to efficiently apply GP regression to a large data set with accurate results. (C) 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.
引用
收藏
页码：703 / 708
页数：6
相关论文
共 15 条
[1]  
Cao Yanshuai., 2013, P 26 ADV NEUR INF PR, V26, P1097
[2]  
Chalupka K, 2013, J MACH LEARN RES, V14, P333
[3]   Sparse on-line Gaussian processes [J].
Csató, L ;
Opper, M .
NEURAL COMPUTATION, 2002, 14 (03) :641-668
[4]   UPDATING THE INVERSE OF A MATRIX [J].
HAGER, WW .
SIAM REVIEW, 1989, 31 (02) :221-239
[5]  
Hensman N., 2013, P 29 C UNC ART INT U, P282, DOI [DOI 10.1162/089976699300016331, DOI 10.48550/ARXIV.1309.6835]
[6]  
Higham N. J., 2002, Accuracy and Stability of Numerical Algorithms, DOI [10.1137/1.9780898718027, DOI 10.1137/1.9780898718027]
[7]   Recursive Gaussian process: On-line regression and learning [J].
Huber, Marco F. .
PATTERN RECOGNITION LETTERS, 2014, 45 :85-91
[8]   Sparse online warped Gaussian process for wind power probabilistic forecasting [J].
Kou, Peng ;
Gao, Feng ;
Guan, Xiaohong .
APPLIED ENERGY, 2013, 108 :410-428
[9]  
Lawrence N.D., 2003, 9 INT WORKSH ART INT
[10]  
Miller K. S., 1981, Mathematics magazine, V54, P67, DOI DOI 10.1080/0025570X.1981.11976898
12