"Predictive Algorithms for Dynamical Sampling Problems with Burst-like Forcing Terms"Krishtal, IlyaWe discuss two algorithms for detecting an unusual occurrence (burst) in an otherwise smooth environment. The key feature of both algorithms is the predictive nature of the design of the samples. The first algorithm utilizes discrete time samples of the environment whereas the second one uses specific weighted averages (Fourier coefficients) of continuous time samples and establishes the time of the burst more accurately. Both algorithms recover the shape of the burst with an error controlled by a measure of smoothness of the environment, measurement acquisition error, and the time step of the algorithm chosen by the user. The talk is based on joint work with A. Aldroubi, L.X. Huang and K. Kornelson. |
« back