Spectral Approach to Solving the Problem of Bithreshold Separability | IEEE Conference Publication | IEEE Xplore

Spectral Approach to Solving the Problem of Bithreshold Separability


Abstract:

The paper deals with the issues concerning the application of methods of harmonic analysis in the theory of bithreshold neural units. We show that the values of spectral ...Show More

Abstract:

The paper deals with the issues concerning the application of methods of harmonic analysis in the theory of bithreshold neural units. We show that the values of spectral coefficients of first two orders of a dichotomy function of the finite set in n-dimensional space give answer to the question of the bithreshold separability of two subsets in which this dichotomy results. The established properties of bithreshold neurons can be useful in the design of off-line learning algorithms developed on the base of the spectral approach.
Date of Conference: 10-12 November 2022
Date Added to IEEE Xplore: 02 January 2023
ISBN Information:

ISSN Information:

Conference Location: Lviv, Ukraine

References

References is not available for this document.