Calculation of derivatives for parameter estimation
A recent letter presents a simple method for the computation of the approximate Hessian Σ{∂εk(θ)/∂θ}{∂εk(θ)/∂θ}T of a cost function of the form ½Σεk2(θ), where θ is an m-dimensional vector of parameters. In this letter, it is pointed out that the method is applicable to more general cost functions of the form ΣLk(θ), e.g. the quadratic form ½ΣvkT(θ) Q-1vk(θ), where vk is an l-dimensional vector. Estimation of the initial state is readily included within the same framework.
See Also
This article has the following associated article(s):
Calculation of derivatives for parameter estimation