Using machine learning algorithms for recursive digital filter optimization
This is an example of using automatic differentiation and optimization methods of TensorFlow to design a 6-order gigital band-pass IIR filter with frequency response as close as possible to the analog filter of the same order.
Initial digital filter is computed using bi-linear transform:
Adam algorithm is used to make the digital filter's frequency response closer to analog filter.