Class-dependent, discrete time-frequency distributions via operator theory

We propose a property for kernel design which results in distributions for each of two classes of signals which maximally separates their energies in the time-frequency plane. Such maximally separated distributions may result in improved classification because the signal representation is optimized to accentuate the differences in signal classes. This is not the case with other time-frequency kernels which are optimized based upon some criteria unrelated to the classification task. Using our operator theory formulation for time-frequency representations, our "maximal separation" criteria takes on a very easily solved form. Analysis of the solution in both the time-frequency and ambiguity planes is given along with an example on discrete signals.

[1]  Moeness G. Amin,et al.  A constrained weighted least squares approach for time-frequency distribution kernel design , 1996, IEEE Trans. Signal Process..

[2]  Robert J. Marks,et al.  The use of cone-shaped kernels for generalized time-frequency representations of nonstationary signals , 1990, IEEE Trans. Acoust. Speech Signal Process..

[3]  Chrysostomos L. Nikias,et al.  A new positive time-frequency distribution , 1994, Proceedings of ICASSP '94. IEEE International Conference on Acoustics, Speech and Signal Processing.

[4]  Patrick Flandrin,et al.  A time-frequency formulation of optimum detection , 1988, IEEE Trans. Acoust. Speech Signal Process..

[5]  I. Vincent,et al.  Non stationary signals classification using time-frequency distributions , 1994, Proceedings of IEEE-SP International Symposium on Time- Frequency and Time-Scale Analysis.

[6]  W. J. Williams,et al.  New properties for discrete, bilinear time-frequency distributions , 1996, Proceedings of Third International Symposium on Time-Frequency and Time-Scale Analysis (TFTS-96).

[7]  W. J. Williams,et al.  An algorithm for positive time-frequency distributions , 1996, Proceedings of Third International Symposium on Time-Frequency and Time-Scale Analysis (TFTS-96).

[8]  Les E. Atlas,et al.  Applications of positive time-frequency distributions to speech processing , 1994, IEEE Trans. Speech Audio Process..

[9]  August W. Rihaczek,et al.  Signal energy distribution in time and frequency , 1968, IEEE Trans. Inf. Theory.

[10]  S. Kay,et al.  On the optimality of the Wigner distribution for detection , 1985, ICASSP '85. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[11]  William J. Williams,et al.  Improved time-frequency representation of multicomponent signals using exponential kernels , 1989, IEEE Trans. Acoust. Speech Signal Process..

[12]  Jasha Droppo,et al.  An operator theory approach to discrete time-frequency distributions , 1996, Proceedings of Third International Symposium on Time-Frequency and Time-Scale Analysis (TFTS-96).

[13]  Patrick J. Loughlin,et al.  An information-theoretic approach to positive time-frequency distributions , 1992, [Proceedings] ICASSP-92: 1992 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[14]  Leon Cohen,et al.  A general approach for obtaining joint representations in signal analysis. I. Characteristic function operator method , 1996, IEEE Trans. Signal Process..

[15]  Les E. Atlas,et al.  Bilinear time-frequency representations: new insights and properties , 1993, IEEE Trans. Signal Process..

[16]  Thomas W. Parks,et al.  Discrete-time, discrete-frequency time-frequency representations , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.