UMLAUT - Little snail

UMLAUT 1.0

Unsupervised Machine Learning Algorithm based on Unbiased Topology (IDL VERSION)

DOI: 323649071.svg

https://zenodo.org/badge/323649071.svg

REFERENCE: Baronchelli I. et al. (2021):

https://iopscience.iop.org/article/10.3847/1538-4365/ac250c/meta

https://arxiv.org/abs/2111.01807

UMLAUT is a variant of the KNN (K-closest neighbor) algorithm.

- Given a set of reference data points (training set), for which the value of N+1 parameters is known,

- Given one analysis data point with N parameters known and the (N+1)-th parameter unknown,

--> UMLAUT estimates the value of the (N+1)-th parameter for the analysis data point. To this purpose, UMLAUT finds the closest data-points of the training set, in a ranked N-dimensional space "associated" with the physical parameter space. After finding the closest data points, the unknown parameter is obtained as the combination (ex. from the average) of the values assumed by the closest reference data points along the (N+1)-th dimension.

Little snail is the first release of the UMLAUT software (Baronchelli, I. et al, 2021) For this initial release:

• only an IDL version (function) is available;
• no particular effort has been made to optimize the code

Ivano Baronchelli, December 10th 2021