|
ROBUST AND SIMPLE ALGORITHMS FOR MAXIMUM LIKELIHOOD ESTIMATION OF MULTIVARIABLE SYSTEMS
Brett Ninness* Stuart Gibson*
* Dept. of Elec. & Comp. Eng, Uni. Newcastle, Australia.
email:brett@ee.newcastle.edu.au, FAX:+61 49 21 69 93
This paper presents novel algorithms for the estimation of dynamic systems. These new methods offer several advantages of being parameterisation free, numerically robust, convergent to statistically optimal estimates, and applicable in a simple fashion to a wide range of multivariable, non-linear and time varying problems. The key tool underlying the new techniques presented here is the Expectation-Maximisation algorithm.
Keywords: Parameter Estimation, Maximum Likelihood Estimators
Session slot T-We-M01: Identification of Linear Systems/Area code 3a : Modelling, Identification and Signal Processing
 | 