Optimal Experimental Design in Canonical Expansions with Applications to Signal Compression
- Additional Document Info
- View All
© 2016 IEEE. In this paper, we introduce an experimental design framework for Karhunen-Loeve compression. This method based on the concept of mean objective of uncertainty determines the best unknown parameter of the covariance matrix to be estimated first in order to improve the quality of the compressed signal. Moreover, we find the closed-form solution to the intrinsically Bayesian robust Karhunen-Loève compression that is required for experimental design and provides the optimal signal compression on average relative to the uncertainty class of covariance matrices. We verify the performance of the proposed experimental design method for the case in which the covariance matrix consists of disjoint blocks.
author list (cited authors)
Dehghannasiri, R., Qian, X., & Dougherty, E. R.