Computer Science – Information Theory
Scientific paper
May 2001
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2001aipc..568..361b&link_type=abstract
BAYESIAN INFERENCE AND MAXIMUM ENTROPY METHODS IN SCIENCE AND ENGINEERING: 20th International Workshop. AIP Conference Proceedi
Computer Science
Information Theory
Data Analysis: Algorithms And Implementation, Data Management, Information Theory And Communication Theory, Telemetry: Remote Control, Remote Sensing, Radar
Scientific paper
Regularization of Autoregressive Spectral Estimation is solved by considering the AR polynomial as a closed planar curve immersed in the complex plane in the framework of Calculus of Variations. This approach allows to recover the classical Regularized Yule-Walker Equation. This formalization is extended by defining a new functional from a new curve length definition in a different Riemannian space where the Euclidean arclength is weighted by the data fitting criteria. The Euler-Lagrange equation provides a steepest-descent method given by a curve evolution equation of the Mean Curvature Flow kind. In this new approach, the hyperparameter of the Lagrangian method has vanished, and has been replaced by a test on speed curve evolution, avoiding any hyperparameter optimization needed in classical Tikhonov technics. We conclude with some generalizations, by considering the autoregressive polynomial as a periodic parametric space curve immersed in CxR. .
No associations
LandOfFree
Calculus of variations and regularized autoregressive spectral estimation does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Calculus of variations and regularized autoregressive spectral estimation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Calculus of variations and regularized autoregressive spectral estimation will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-924061