Two algorithms for designing optimal reduced-bias data-driven time-frequency detectors
Abstract
Designing time-frequency detectors from training data is potentially of great benefit when few a priori information on the non-stationary signal to be detected is available. However, achieving good performance with data-driven detectors requires matching their complexity to the available amount of training samples: receivers with a too large number of adjustable parameters often exhibit poor generalization performance whereas those with an insufficient complexity cannot learn all the information available in the set of training data. We present two methods which provide powerful tools for tuning the complexity of time-frequency detectors and improving their performance. These procedures may offer an helpful support for designing efficient detectors from small training sets, in applications of current interest such as biomedical engineering and complex systems monitoring.