Lazy Gaussian Process Committee for Real-Time Online Regression

Abstract

A significant problem of Gaussian process (GP) is its unfavorable scaling with a large amount of data. To overcome this issue, we present a novel GP approximation scheme for online regression. Our model is based on a combination of multiple GPs with random hyperparameters. The model is trained by incrementally allocating new examples to a selected subset of GPs. The selection is carried out efficiently by optimizing a submodular function. Experiments on real-world data sets showed that our method outperforms existing online GP regression methods in both accuracy and efficiency. The applicability of the proposed method is demonstrated by the mouse-trajectory prediction in an Internet banking scenario.

Cite

Text

Xiao and Eckert. "Lazy Gaussian Process Committee for Real-Time Online Regression." AAAI Conference on Artificial Intelligence, 2013. doi:10.1609/AAAI.V27I1.8572

Markdown

[Xiao and Eckert. "Lazy Gaussian Process Committee for Real-Time Online Regression." AAAI Conference on Artificial Intelligence, 2013.](https://mlanthology.org/aaai/2013/xiao2013aaai-lazy/) doi:10.1609/AAAI.V27I1.8572

BibTeX

@inproceedings{xiao2013aaai-lazy,
  title     = {{Lazy Gaussian Process Committee for Real-Time Online Regression}},
  author    = {Xiao, Han and Eckert, Claudia},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2013},
  pages     = {969-976},
  doi       = {10.1609/AAAI.V27I1.8572},
  url       = {https://mlanthology.org/aaai/2013/xiao2013aaai-lazy/}
}