pp_wavelet_features — Multilevel wavelet summary features¶
Group: Augmentation · C ABI: n4m_transform_wavelet_features_*
Description¶
Multi-level DWT summary features.
Parameters¶
Name |
Type |
Default |
|---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
API and bindings¶
C ABI (ABI 2): n4m_transform_wavelet_features_create · n4m_transform_wavelet_features_create_ex · n4m_transform_wavelet_features_destroy · n4m_transform_wavelet_features_output_cols · n4m_transform_wavelet_features_transform. Use the linked public header for the exact signature, configuration, and result handles.
Python (verified public re-export):
from n4m.transform.wavelet import WaveletFeatures
Source signature: WaveletFeatures(family: str = 'haar', mode: str = 'periodization', max_level: int = 3, entropy: str = 'energy').
R: no current source-verified entry point was found for this catalog method.
MATLAB / Octave: no current source-verified entry point was found for this catalog method.
Explanations¶
Bibliographic source¶
No single canonical paper defines this exact four-statistic descriptor. It uses the multiresolution analysis of Mallat (1989), https://doi.org/10.1109/34.192463, with implementation-specific summaries.
Mathematical principle¶
After multilevel DWT, every approximation/detail band contributes four values: mean, population standard deviation, energy \(\sum c_i^2\), and either normalized-energy entropy or ten-bin histogram entropy. Output width is \(4(L+1)\).
Appropriate uses¶
Producing compact, fixed-size multiscale descriptors for classification or regression.
Limits and validation¶
Band summaries discard coefficient position and sign structure beyond the mean. Entropy definitions are implementation-specific and depend on family, mode, and feasible decomposition depth.
Implementation¶
n4m.transform.wavelet.WaveletFeatures uses n4m_transform_wavelet_features_*; all four summaries and entropy modes are in preprocessing/wavelets/wavelet_features.c.
The ABI-2 implementation is the n4m_transform_wavelet_features_* lifecycle in libn4m.
Sources and provenance¶
https://doi.org/10.1109/34.192463; https://github.com/GBeurier/nirs4all-methods/blob/main/cpp/src/core/preprocessing/wavelets/wavelet_features.c
See also: methods index · interactive dashboard