The field of Automated Learning (Machine Learning) is a central part of the new technological revolution based on the intelligent use of information. Traditionally, the main problems that are investigated in this area are those of pattern recognition or classification, function approximation or regression of continuous variables, and the discovery of hidden structures in data or Clustering. Logically, the development of new methods and algorithms are focused at first on the most simple and typical ones to find, for example in stationary problems in time, with an abundance of examples to learn from and with only a few classes fairly balanced between them. However, new types of data from genomics, proteomics and continuous monitoring equipment of critical systems have introduced new challenges in Automated Learning. This project proposes the development of new methods, or the extension of existing techniques when appropriate, to efficiently model this new kind of data, including unsteady regression classification problems or with great noise, clustering classification problems with an extremely high number of input variables or classification problems with a significant imbalance between classes. All lines of this project include applications to current issues of great technological interest, covering areas of biotechnology, agriculture and steel industries.
Dr. Pablo Granitto
Investigador Principal de CONICET y Profesor Titular de UNR
Investigate in Aprendizaje automatizado (machine learning). Métodos de ensemble. Selección de características. Clustering.