Code for the Multiscale kernel regression, available under the license CeCILL-B
Release 1.0 = MATLAB / Author: Nicolas Duchateau (CREATIS Lyon, France) / February 2018
Release 2.0 = PYTHON / Authors: Benoit Freiche, Fei Zheng, Nicolas Duchateau (CREATIS Lyon, France) / August 2023
Links to the corresponding publications at:
https://www.creatis.insa-lyon.fr/~duchateau/#publications
NOTICE:
This code is made open-access. Comments and bug reports are welcome, as well as feedback on its possible improvements.
Published reports of research using this code (or a modified version) may cite the following articles that describes the method:
- exact matching case:
Bermanis A, Averbuch A, Coifman RR. Multiscale data sampling and function extension. Applied and Computational Harmonic Analysis, 2013;34(1):15-29. https://doi.org/10.1016/j.acha.2012.03.002
- inexact matching case, corresponding to the present MATLAB implementation:
Duchateau N, De Craene M, Sitges M, Caselles V. Adaptation of multiscale function extension to inexact matching: Application to the mapping of individuals to a learnt manifold. In: Proceedings of SEE International Conference on Geometric Science of Information (GSI). Springer LNCS, 2013;8085:578-86. https://doi.org/10.1007/978-3-642-40020-9_64
ARCHIVE CONTENT [MATLAB]
testSIN.m = example to launch the code on a sinusoid at different frequencies (to test the regression at different scales)
interpolateMulti.m = launching the multi-scale routine (algorithm #4 from adapted Bermanis et al. 2013)
MINEXACT_Bermanis_ACHA_2013_ALGO_03.m = computations at a single scale (algorithm #3 adapted from Bermanis et al. 2013)
ARCHIVE CONTENT [PYTHON]
regression.py = class MultiScaleKernelRidge + example to launch the code on a sinusoid at different frequencies (to test the regression at different scales)