Lecture: 'Some insights on explainability & Read Write Machine Learning'

This is a short presentation of LIAAD, the lab of Artificial Intelligence and Decision Support, a centre of INESC TEC (https://www.inesctec.pt/en). Two different topics will be presented:

1. Works related to explainability:
   • On merging decision trees as a way to obtain interpretable profiles of a given attribute of interest.
   • Giving personalized explanations using counteRGAN (counterfactuals with GANs).
   • Using background knowledge as a way to give explainability in the context of online event detection.
2. Read & Write Machine Learning (rwML):
   • A new paradigm, or on how to not only read the data in the training phase to improve predictions. cLSTM, an example of an rwML method.

Bio

João Mendes Moreira received his PhD degree in Engineering Sciences from the Faculty of Engineering of the Porto University (FEUP) in 2008. He is an Associate Professor in the Department of Informatics Engineering, FEUP, and a senior researcher at LIAAD (Laboratory for Artificial Intelligence and Decision Aid), a centre belonging to INESC TEC. He has worked on projects and authored papers in areas that are mainly related to the application of machine learning to real-world problems in the areas of intelligent transportation systems, precision agriculture, human activity recognition, among others. His area of specialization is supervised learning with a special interest in spatio-temporal learning, ensemble learning, outlier/novelty detection, imbalanced datasets, all these using both batch and online learning. He is the deputy director of the master in data science and engineering at FEUP. He is co-author of the Wiley’s book “A general introduction to data analytics”, 2019.