The two are the co-authors of the paper on deep learning, a part of artificial intelligence that deals with emulating the learning approach used by humans. This paper was awarded second prize for best paper among the hundred or so submitted at this scientific meeting of leading world specialists.
In their work the authors proposed modifying the way in which information is merged in specific phases of the machine learning process with deep neural networks (deep learning). "Until now," explained Prof Humberto Bustince, "this merging has been conducted using operators such as the maximum or the mean. The work proposes using a type of integral known as the Choquet integral which, through the use of appropriate measures, allows the existing relationships between the data to be considered. This modification improves the performance of deep neural networks in classification problems, in particular in the case of images".
###
The two NUP/UPNA lecturers, assigned to the Department of Statistics, Computing and Mathematics and also to Navarrabiomed (biomedical research centre of the Government of Navarre and the NUP/UPNA) and the Institute of Health Research of Navarre (IdiSNA), wrote the award-winning paper in collaboration with researchers at the Federal University of Río Grande (Brazil).