A note on AIC-type criteria, MDL principle and nonlinear model selection

Huang, Fuchun (2004) A note on AIC-type criteria, MDL principle and nonlinear model selection. In: Proceedings of The International Conference on Optimization: Techniques and Applications, 9 December 2004, Ballarat, Australia. (Unpublished)

Full text not available from this repository.

Abstract

By two examples this note illustrates that in nonlinear model selection it is essentially not the number of parameters but the total description length of the model, including the description length of the parameters, should be penalized in cooperating with the `goodness-of-¯t' of the model to the data. The fact illustrated is in favor of MDL principle rather than AIC-type criteria in nonlinear model selection.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	AIC-type criteria, MDL principle, nonlinear regression, nonlinear autoregression, model selection
Subjects:	RFCD Classification > 280000 Information, Computing and Communication Sciences School/Research Centre/Department > School of Computer Science and Mathematics
ID Code:	1549
Deposited By:	Ms Phung T Tran
Deposited On:	31 Oct 2008 08:51
Last Modified:	31 Oct 2008 08:51
ePrint Statistics:	View statistics for this ePrint

Repository Staff Only: item control page