Propagation of uncertainty for an Adaptive Linear Approximation algorithm

Abstract

In machine learning, a variety of algorithms are available for feature extraction. To obtain reliable features from measured data, the propagation of measurement uncertainty is proposed here in line with the Guide to the Expression of Uncertainty in Measurement (GUM). Recently, methods for the discrete Fourier and Wavelet transform have been published. Here, the Adaptive Linear Approximation (ALA) as a further complementary feature extraction algorithm is considered in combination with an analytical model for the uncertainty evaluation of the ALA features.