We propose two modifications of Gaussian processes, which aim to deal with dynamic environments. One is a weight decay method that gradually forgets old data, and the other is a time stamp method that regards the time course of data as a Gaussian process. We show experimental results when these modifications are applied to regression problems in dynamic environments. The weight decay method is found to follow the environmental change by automatically ignoring the past data, and the time stamp method is found to predict linear alteration.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Shigeyuki OBA, Masa-aki SATO, Shin ISHII, "On-Line Learning Methods for Gaussian Processes" in IEICE TRANSACTIONS on Information,
vol. E86-D, no. 3, pp. 650-654, March 2003, doi: .
Abstract: We propose two modifications of Gaussian processes, which aim to deal with dynamic environments. One is a weight decay method that gradually forgets old data, and the other is a time stamp method that regards the time course of data as a Gaussian process. We show experimental results when these modifications are applied to regression problems in dynamic environments. The weight decay method is found to follow the environmental change by automatically ignoring the past data, and the time stamp method is found to predict linear alteration.
URL: https://global.ieice.org/en_transactions/information/10.1587/e86-d_3_650/_p
Copy
@ARTICLE{e86-d_3_650,
author={Shigeyuki OBA, Masa-aki SATO, Shin ISHII, },
journal={IEICE TRANSACTIONS on Information},
title={On-Line Learning Methods for Gaussian Processes},
year={2003},
volume={E86-D},
number={3},
pages={650-654},
abstract={We propose two modifications of Gaussian processes, which aim to deal with dynamic environments. One is a weight decay method that gradually forgets old data, and the other is a time stamp method that regards the time course of data as a Gaussian process. We show experimental results when these modifications are applied to regression problems in dynamic environments. The weight decay method is found to follow the environmental change by automatically ignoring the past data, and the time stamp method is found to predict linear alteration.},
keywords={},
doi={},
ISSN={},
month={March},}
Copy
TY - JOUR
TI - On-Line Learning Methods for Gaussian Processes
T2 - IEICE TRANSACTIONS on Information
SP - 650
EP - 654
AU - Shigeyuki OBA
AU - Masa-aki SATO
AU - Shin ISHII
PY - 2003
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E86-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2003
AB - We propose two modifications of Gaussian processes, which aim to deal with dynamic environments. One is a weight decay method that gradually forgets old data, and the other is a time stamp method that regards the time course of data as a Gaussian process. We show experimental results when these modifications are applied to regression problems in dynamic environments. The weight decay method is found to follow the environmental change by automatically ignoring the past data, and the time stamp method is found to predict linear alteration.
ER -