Learning contextual information via deep learning

Abstract

[en] During the last few years, deep learning has become one of the most attractive fields of artificial intelligence, with the use of artificial neural networks at its core. In this project we propose several neural networks architectures for the context learning methodology. The main goal of this project is to verify if these methodologies might work on medical images by first testing them on simpler datasets. We propose two different approaches, one consisting of a convolutional architecture and the other being a recurrent neural network. Whilst the first approach provided grate results with the first datasets we used, it proved to be insufficient as the complexity of the dataset increased. The recurrent architecture provided successful results when working with more complex datasets. This thesis provides a general overview of neural networks and explains the different steps taken to reach the proposed models.