A mixture-of-experts framework for adaptive Kalman filtering

Abstract

Abstract-This paper proposes a modular and flexible approach to adaptive Kalman filtering using the framework of a mixture-of-experts regulated by a gating network. Each expert is a Kalman filter modeled with a different realization of the unknown system parameters such as process and measurement noise. The gating network performs on-line adaptation of the weights given to individual filter estimates based on performance. This scheme compares very favorably with the classical Magill filter bank, which is based on a Bayesian technique, in terms of i) estimation accuracy, ii) quicker response to changing environments, and iii) numerical stability and computational demands. The proposed filter bank is further enhanced by periodically using a search algorithm in a feedback loop. Two search algorithms are considered. The first algorithm uses a recursive quadratic programming approach which extremizes a modified maximum likelihood function to update the parameters of the best performing filter in the bank. This particular approach to parameter adaptation allows a real-time implementation. The second algorithm uses a genetic algorithm to search for the parameter vector and is suited for post-processed data type applications. The workings and power of the overall filter bank and the suggested adaptation schemes are illustrated by a number of examples.

Mixture kernel least mean square

“…It has been popularly used to combine models learned with different realizations of unknown system parameters such as process and measurement noise [17], [18], [15] or different subsets of input data [19], in order to address issues related to non-stationarity, accuracy, computational burden, etc. The common aspect on these methods and our method is that, the experts compete to represent the input at each time instance.…”

Superior foetal electrocardiogram signal elicitation using a novel artificial intelligent Bayesian methodology

“…The conventional KF utilizes a linear replica for the system dynamics and mathematical observations. However, most practical systems are nonlinear in nature and in order to broaden the scheme of conventional KF to such nonlinear systems, numerous variants of the original KF methodologies and add-ons have been developed [17,18]. The extended Kalman filter (EKF) is one amongst the various techniques of the conventional Kalman Filtering for applications to nonlinear systems [8] such as the signals that are being dealt with, in this work.…”

Research on two different mathematical theories on control

“…In the past, the probability statistics (PS) method [11,[22][23][24][25] was usually used for target control. Along with the development of information and industry technology, at present, people are becoming more and more interested in fuzzy control [1,[3][4][5][6][7][8][9][10]12,13,15,[17][18][19]21,26] and rough sets for target control [2,20,27,28]; again, by the combination of fuzzy sets (FS) theory and rough sets (RS) theory, i.e., fuzzy rough sets (FRS) theory [14,16,29], we can obtain a new control algorithm, which here is called the fuzzy rough (FR) control algorithm.…”