Code to use learnable wavelet transforms like L-WPT and DeSPaWN methods in pytorch
Implementation of the L-WPT [1], [2] and Despawn (L-DWT) [3]. L-WPT is a learnable extension of the Wavelet Packet Transform while Despawn (L-DWT) is a learnable extension of the Discrete Wavelet Transform. Both architecture can be called using the class NeuralDWAV.py (Neural Discrete Wavelet). Read me in progress
[1] Frusque, G., & Fink, O. (2022). Robust time series denoising with learnable wavelet packet transform. arXiv preprint arXiv:2206.06126.
[2] Frusque, G., & Fink, O. (2022, May). Learnable wavelet packet transform for data-adapted spectrograms. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 3119-3123). IEEE.
[3] Michau, G., Frusque, G., & Fink, O. (2022). Fully learnable deep wavelet transform for unsupervised monitoring of high-frequency time series. Proceedings of the National Academy of Sciences, 119(8), e2106598119.