This thesis is focused on exploring explainable AI algorithms and in particular Layer-Wise Relevance Propagation (LRP). We explore different LRP rules and analyze the values inside the network to understand the saliency ...

This work propose a methodology for still image facial expression. The proposed method contains a face detection and alignment module followed by a deep convolutional neural network (CNN) that outputs a seven emotions ...

We evaluate the robustness of reward functions learned with IRL, when transferred to similar tasks. We exceed state of the art results for one benchmark task and solve another one for the first time. Modifications are ...

The aim of this thesis is to apply multiresolution to Infomap, a community detection algorithm based in the map equation that work in monolayer and multilayer networks.

We consider a case of covariate shift where prior causal inference or expert knowledge has identified some features as effects, and show how this setting, when analysed from a causal perspective, gives rise to a semi-generative ...

The Mediterranean city of Tarragona was a capital city in Roman times. Nowadays, the remains of the ancient constructions coexist with many subsequent buildings. This is the case of the Gothic Cathedral cloister, whose ...