Cookies Policy
The website need some cookies and similar means to function. If you permit us, we will use those means to collect data on your visits for aggregated statistics to improve our service. Find out More
Accept Reject
  • Menu
Article

INESC TEC paper on convolutional neural networks was distinguished as «Best Paper»

«Producing Decisions and Explanations: A Joint Approach Towards Explainable CNNs» is the title of the paper, involving two INESC TEC researchers, which deserved the distinction of BEST PAPER at RECPAD – Annual Conference on pattern recognition, that took place on 31 October at the Faculty of Sciences of the University of Porto.

30th December 2019

«Producing Decisions and Explanations: A Joint Approach Towards Explainable CNNs» is the title of the paper, involving two INESC TEC researchers, which deserved the distinction of BEST PAPER at RECPAD – Annual Conference on pattern recognition, that took place on 31 October at the Faculty of Sciences of the University of Porto.

At issue is the research, carried out by Kelwin Fernandes and Luís Teixeira, of the Centre for Telecommunications and Multimedia (CTM), who, together with the researcher Isabel Rio-Torto, introduced a model for convolutional neural networks (CNN – Convolutional Neural Networks). The goal is to implement a digital image reading system, independent from human intervention, through a CNN architecture composed of an explainer and a classifier.

The results of the paper showed that the new CNN architecture proposed in the paper is capable of producing explanations of images, without the need for supervision, as well as decisions, while retaining the accuracy in the classification.  These advances in CNN can potentially be applied to several activities, with emphasis on medical image, image recognition and video processing, among others.

Convolutional neural networks are inspired in biological processes, more specifically in the visual cortex of animals and the way it processes images.

 

The researchers mentioned in this news piece are associated with UP-FEUP.