Pamela Zontone
Assistant Professor, University of Genoa
Bio:
Pamela Zontone earned a Laurea in Electronic Engineering from the University of Udine in 2004, and a PhD in Information and Industrial Engineering from the University of Udine in 2008. From 2009 to 2011, she was a postdoctoral fellow at the Department of Information Engineering and Computer Science, University of Trento, working on the LivingKnowledge European project. In 2017, she joined the Polytechnic Department of Engineering and Architecture, University of Udine, where she worked in the field of sensor signal processing and machine learning techniques applied to biophysical signals. She is currently an assistant professor (RTD-a) at the Department of Electrical, Electronic, Telecommunications Engineering and Naval Architecture, University of Genoa. Her research interests include multidimensional signal processing, biophysical signal processing, and machine learning.
Workshop:
Advanced sensors and machine learning techniques for biomedical signal processing
with Antonio Affanni
Abstract:
Stress, anxiety, drowsiness, and other emotional states are important factors that should be assessed in car drivers. These states can affect their physical and physiological well-being and, as a result, can influence their behaviour and reactions to potentially dangerous road situations.
The first part of the lecture will show an overview of commercial wearable sensors used in literature to classify and detect mental states of drivers. In particular, we will explore the possible metrological issues that arise when using commercial devices and we will show the advantages/drawbacks of using commercial sensors versus ad hoc developed prototypes. Since the sensors are the first link in the chain of machine learning and deep learning algorithms, their metrological characterization and traceability is of paramount importance in order to avoid misleading results in classification.
In the second part of the lecture, we will cover the fundamentals of biosignals and machine learning (ML) algorithms. Additionally, we will introduce some of the ML architectures we have designed over the years for stress detection in car drivers. Specifically, these systems are based on the analysis of physiological signals, such as the electrodermal activity (EDA), electrocardiogram (ECG), and electroencephalogram (EEG) signals, to detect the different emotional states experienced by subjects while driving in various simulated scenarios. We will discuss the results obtained from experiments carried out during both manual and autonomous driving scenarios, as well as while driving in a simulated urban area characterized by various traffic situations.
