M/511

We describe the construction of frames where the coefficients are based on Fourier coefficients of a function, and characterize local smoothness of the function on different intervals. The ideas are used for blind source signal separation problems, chirp detection in particular. We present some results treating the classification problems in machine learning as a generalization of the signal separation problem.

Prof. Dr. Hrushikesh Mhaskar

Claremont Graduate University, USA