MaLGa logoMaLGa black extendedMaLGa white extendedUniGe ¦ MaLGaUniGe ¦ MaLGaUniversita di Genova | MaLGaUniversita di Genova

Possible directions to deal with EEG data with machine learning




Possible directions to deal with EEG data with machine learning


Andrea Apicella - Università di Napoli Federico II


Brain-computer interface (BCI) systems based on EEG signals are gaining increasing attention in applications due to the noninvasiveness and high temporal resolution of EEG acquisition devices. However, the nonstationarity of EEG signals makes the use of the systems susceptible to differences between users or temporal differences between acquisitions. This problem can be seen as an instance of the well-known dataset shift problem in Machine Learning. This talk presents and discusses the methods and future directions explored in our lab to address the dataset bias problem in the context of EEG-based BCIs. In particular, simple data normalization strategies' role in EEG classification performance is discussed. A preliminary analysis of the use of Explicit Artificial Intelligence (XAI) to find relevant features of input EEG data is then presented, suggesting a new direction toward new BCI methods that may achieve better generalization performance.


Andrea Apicella received the M.Sc. degree in Computer Science and the Ph.D. degree in Mathematics and Computer Science from the Federico II University of Naples, Italy, in 2014 and 2019, respectively. He is currently a Research Fellow at the Department of Information Technology and Electrical Engineering, University of Naples Federico II. He is also a Research Associate at ARHeMLab (Augmented Reality for Health Monitoring Laboratory) and the AIPA Lab (Laboratory of Artificial Intelligence, Privacy and Applications), DIETI Excellence Department, University of Naples Federico II. His current research interests include EEG signal processing for emotion recognition, attention monitoring using artificial intelligence methods, and eXplainable artificial intelligence (XAI) approaches for explaining the AI system’s decisions.


December 19th 2022, 11:30


Room 710, UniGe DIMA, Via Dodecaneso 35