Context:

In engineering, modeling for risk analysis and ensuring the validity of results in systems that include computational routines requires the analysis of the sources and categories of uncertainty, which, in this context, can be classified as aleatoric and epistemic.

Method:

A literature review from databases such as Google Scholar, IEEEXplore, and ScienceDirect is presented herein, analyzing trends and approaches related to the concept of uncertainty within the framework of soft metrology, in order to improve our understanding when there are additional restrictions due to the assurance of the validity of the results.

Results:

This paper presents concepts and comparisons that aid in the understanding of epistemic and random uncertainty in soft metrology measurement processes and the way in which it is related to the assurance of the validity of results within the framework of learning machines.

Conclusions:

Representation quality in soft metrology systems is constantly influenced by random uncertainty, while epistemic uncertainty exhibits descending dynamics when the fit of the model improves with sufficient training data.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )