Explicability, Privacy and Equity in Machine Learning
Artificial Intelligence (AI) offers increasingly precise algorithms that make it possible to analyse all kinds of data or signals, such as time series, images, text data, etc.
With the emergence of applications, new needs have arisen which, along with precision, are fixed by the AI applications themselves and the problems they address. Among them, there is the explanation of the decision making of the algorithms, algorithms that we can audit, with follow-up to the traceability of their behaviour, as well as avoiding that discriminations are produced in their use.
On the other hand, data privacy is essential in many applications in which data must be preserved in the centres of origin of the data, so algorithms are required that can extract knowledge presenting privacy.
In this line of research we are considering the development of artificial intelligence and learning algorithms that preserve privacy, avoiding discrimination caused by the bias of the data (fairness), auditable and transparent. As summary, we discuss about privacy and FATE (Fairness, Accountability, Transparency and Ethics).
Contact: Francisco Herrera Triguero