ITI develops a tool to detect the deterioration of Artificial Intelligence models

VALENCIA, 28 Mar. (EUROPA PRESS) -

The ITI technology center, specialized in ICT, has developed AITANA-MoMo, an innovative project that monitors and detects the degradation of Artificial Intelligence models. This initiative will enable developers to make decisions about the models put into production, as well as investigate the cause of errors and re-train them if necessary, the entity has reported in a statement.

AITANA-MoMo aims to control the factors that can alter the performance of Artificial Intelligence tools, as well as completing "the crucial and important need" in the life cycle of these models, explained ITI research scientist David Millán. .

The AITANA-MoMo initiative, financed by IVACE and FEDER funds, has established specific objectives to develop these AI tools, including carrying out an analysis that allows the identification of variables; the design of a set of functionalities for monitoring; or the implementation of the graphical interfaces necessary to support the new functionalities developed. These objectives are based on two key pillars: the analysis of the status of monitoring algorithms and the definition of KPI's variables.

The functionality of AITANA-MoMo will be put into practice with the creation of a demonstrative pilot test that allows ICT companies in the Valencian Community to test the new functionalities developed in the project in an environment that is as realistic as possible.

"The application will have a significant economic impact and, given the type of companies that will be able to benefit from it, it is essential to carry out a joint demonstration so that they become familiar with the tool", stated the researcher.

Through AITANA-MoMo, the aim is to obtain a detailed study of the techniques and processes that are carried out to be able to detect the degradation of the models, as well as to respond to the enormous number of libraries, tools and development languages, where the Data scientists encounter difficulties because existing solutions address partial problems and do not have the necessary flexibility.

"This tool will make models available to specialists that allow various ways of working, increasing their productivity and comfort", Millán added.

Finally, the AITANA-MoMo project, with file IMDEEA/2022/50, has been co-financed by the Valencian Institute of Business Competitiveness (IVACE) and the European Regional Development Fund (FEDER).