Neural network model of wavelets for the continuous wavelet transform

The existing methods of wavelet synthesis for CWT either do not provide an analytic representation of the wavelet function, or lead to a significant deviation of the synthesized wavelet from the original fragment. The proposed by the author neural network wavelet models allow to avoid both disadvantages. Synthesized wavelets correspond to the admissibility conditions and can be used both in the direct continuous wavelet transform and the reverse continuous wavelet transform. This is confirmed by the simulation test results in MATLAB.