November 2019
Intermediate to advanced
296 pages
7h 52m
English
The implementation that TensorFlow.js uses for Fourier transform is known as FFT. Since the naive implementation of DFT is slow in general use cases, FFT is popularly used in the practical field. The time complexity of DFT in terms of matrix multiplication is
, where N is the data size. FFT reduces the order of the computation to
. This is a huge difference, especially for long sequences. The benefit of FFT is not only its speed but also its accuracy since it removes rounded-off errors. There are many derived family implementations ...
Read now
Unlock full access