AITANA: Standardized working environment for Machine Learning and Recognition projects

The AITANA project aims to develop a set of tools including a graphical user interface to facilitate the development, experimentation and production deployment of Machine Learning and Artificial Intelligence algorithms with complete integration of third-party libraries.

In the frame of projects related to machine learning and pattern recognition, the whole life cycle of a project has the following phases: gathering data, storage, cleaning, homogenization, data integration, training, evaluation and deploying of the models to production.

This requires an extensive knowledge from the data scientists, because there are a huge number of libraries, programming languages and software in general that cover different parts of the life cycle. In these points are where the scientists of data find the early difficulties, since the available solutions partially address the life cycle or these requires an high level of knowledge or the libraries do not have the necessary flexibility.

AITANA covers the complete development life cycle of this kind of application. AITANA covers from the initial analysis, training in computational nodes, to the final step, deploying to production. In this way, the work and research of the data scientist are facilitated. This document summarizes the main characteristics of the AITANA framework.