TY - GEN
T1 - Parameter estimation for Relational Kalman Filtering
AU - Choi, Jaesik
AU - Amir, Eyal
AU - Xu, Tianfang
AU - Valocchi, Albert J.
N1 - Funding Information:
This work is supported by the National Science Foundation Hydrologic Science Program under Grant No. 0943627. This wor was supported by the year of 2014 Research Fund of UNIST (Ulsan National Institute of Science and Technology)
Publisher Copyright:
© Copyright 2014 Association for the Advancement of Artificial Intelligence. All rights reserved.
PY - 2014
Y1 - 2014
N2 - The Kalman Filter (KF) is pervasively used to control a vast array of consumer, health and defense products. By grouping sets of symmetric state variables, the Relational Kalman Filter (RKF) enables to scale the exact KF for large-scale dynamic systems. In this paper, we provide a parameter learning algorithm for RKF, and a regrouping algorithm that prevents the degeneration of the relational structure for efficient filtering. The proposed algorithms significantly expand the applicability of the RKFs by solving the following questions: (1) how to learn parameters for RKF in partial observations; and (2) how to regroup the degenerated state variables by noisy real-world observations. We show that our new algorithms improve the efficiency of filtering the large-scale dynamic system.
AB - The Kalman Filter (KF) is pervasively used to control a vast array of consumer, health and defense products. By grouping sets of symmetric state variables, the Relational Kalman Filter (RKF) enables to scale the exact KF for large-scale dynamic systems. In this paper, we provide a parameter learning algorithm for RKF, and a regrouping algorithm that prevents the degeneration of the relational structure for efficient filtering. The proposed algorithms significantly expand the applicability of the RKFs by solving the following questions: (1) how to learn parameters for RKF in partial observations; and (2) how to regroup the degenerated state variables by noisy real-world observations. We show that our new algorithms improve the efficiency of filtering the large-scale dynamic system.
UR - http://www.scopus.com/inward/record.url?scp=84974827226&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84974827226&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84974827226
T3 - AAAI Workshop - Technical Report
SP - 22
EP - 28
BT - Statistical Relational AI - Papers Presented at the 28th AAAI Conference on Artificial Intelligence, Technical Report
PB - AI Access Foundation
T2 - 28th AAAI Conference on Artificial Intelligence, AAAI 2014
Y2 - 28 July 2014
ER -