A geometrical interpretation of exponentially embedded families of gaussian probability density functions for model selection
Document Type
Article
Date of Original Version
1-1-2013
Abstract
Model selection via exponentially embedded families (EEF) of probabilitymodels has been shown to perform well onmany practical problems of interest. A key component in utilizing this approach is the definition of a model origin (i.e. null hypothesis) which is embedded individually within each competingmodel. In this correspondence we give a geometrical interpretation of the EEF and study the sensitivity of the EEF approach to the choice of model origin in a Gaussian hypothesis testing framework. We introduce the information center (I-center) of competing models as an origin in this procedure and compare this to using the standard null hypothesis. Finally we derive optimality conditions for which the EEF using I-center achieves optimal performance in the Gaussian hypothesis testing framework.© 2012 IEEE.
Publication Title, e.g., Journal
IEEE Transactions on Signal Processing
Volume
61
Issue
1
Citation/Publisher Attribution
Costa, Russell, and Steven Kay. "A geometrical interpretation of exponentially embedded families of gaussian probability density functions for model selection." IEEE Transactions on Signal Processing 61, 1 (2013): 62-67. doi: 10.1109/TSP.2012.2222393.