Enhancing Trustworthiness in Automated Driving through NeuroSymbolic AI
Details
Abstract: This talk explores enhancing the trustworthiness of automated driving systems through NeuroSymbolic AI. This talk propose methods where symbolic representations of qualitative spatial and temporal relationships between objects are learned from video scenes. These symbolic representations are then used as input for machine learning and pattern-mining approaches for understanding complex driving scenarios. This approach enables the generation of explanations for automated driving actions, facilitates the creation of comprehensive testing scenarios, and ultimately contributes to a more transparent and trustworthy autonomous system.
Speaker: Nassim Belmecheri, PhD, is a Research Scientist at SIMULA Research Laboratory in Oslo, Norway. He earned his Ph.D. in Computer Science from the University Oran 1 in July 2023, with his thesis focusing on "Web Data Mining: Learning from User Feedback". Prior to his current role, he served as a Postdoctoral Researcher at SIMULA. Dr. Belmecheri's expertise is at the intersection of Artificial Intelligence (AI), Data Mining, Explainable AI, and Constraint Programming. His professional experience includes developing and deploying AI solutions for companies like AXLR SATT , and working as a Data Scientist with geo-spatial data and text analysis at EvalAndGo/Qwesteo. The past 3 years he has worked on enhancing trustworthiness of AI systems in automated driving through neurosymbolic AI in the context of a European project (AI4CCAM).