Trainable Wavelet Neural Network for Non-Stationary Signals
This work addresses digital signal processing for non-stationary signals, offering a novel method that is incremental in combining wavelets with neural networks.
The authors tackled the problem of processing non-stationary signals by introducing a wavelet neural network that learns a specialized filter-bank, resulting in improved interpretability and performance, with experimental results showing it outperforms standard architectures and generalizes well on noisy data.
This work introduces a wavelet neural network to learn a filter-bank specialized to fit non-stationary signals and improve interpretability and performance for digital signal processing. The network uses a wavelet transform as the first layer of a neural network where the convolution is a parameterized function of the complex Morlet wavelet. Experimental results, on both simplified data and atmospheric gravity waves, show the network is quick to converge, generalizes well on noisy data, and outperforms standard network architectures.