EVENT

Event News

Talk on "The "AI Knowledge" Project - Neuro-symbolic AI in Autonomous Driving" by Prof. Dr. rer. nat. Adrian Paschke from Freie Universität Berlin

We are pleased to inform you about the upcoming seminar by Prof. Dr. rer. nat. Adrian Paschke from Freie Universität Berlin titled : "The "AI Knowledge" Project - Neuro-symbolic AI in Autonomous Driving" Everyone interested is cordially invited to attend!

Title:

The "AI Knowledge" Project - Neuro-symbolic AI in Autonomous Driving

Abstract:

The "AI Knowledge" Project - Neuro-symbolic AI in Autonomous Driving Description: AI-based processes and methods are paving the way to fully automated autonomous driving (AD). Data driven machine learning (ML) approaches for AD require enormous amounts of data for the training and validation of AI functions, with the collection and processing of this data being very resource-intensive and expensive. In addition to the dependence on extensive amounts of data, data-based AI processes have another weakness: they are still generally black-box models for which the decision-making process cannot be directly reconstructed. In my talk I will report about the project "AI Knowledge" (https://www.kiwissen.de/) and the RECOMP project (https://research.nii.ac.jp/RECOMP/), which aims to enhance reliability of norm-based AI by implementing a real-time compliance mechanism for legal and ethical norms. The goal of the AI Knowledge project is to create a comprehensive ecosystem for the integration of knowledge into the training and safeguarding of AI functions. It develops neuro-symbolic AI methods for integrating existing knowledge into the data-driven AI functions of autonomous vehicles (AV) and vice versa for extracting interpretable symbolic knowledge from deep neural network models. By combining conventional data-based AI methods with the knowledge- and rule-based methods developed in the project, the basis for training and validating of AI functions will be completely redefined: The development from data- to knowledge-based neuro-symbolic semantic AI carried out in the project addresses the central challenges towards autonomous driving: the generalization of AI to phenomena with small data bases, the increase of the stability of the trained AI to disturbances in the data, the data efficiency, the plausibility check and the validation of AI-supported functions as well as the increase of the functional quality. Furthermore, recent advantages in generative AI and large language models have facilitated the development of agent-based systems. Despite their encouraging results in various reasoning tasks, these systems often operate as "black boxes", raising concerns about potential illegal behaviour due to opaque decision-making processes. This concern is particularly critical in autonomous driving, where precise decision-making requires a thorough understanding of traffic scenes and strict adherence to established norms. In this talk, I will introduce a legally-guided automated decision making system that employs language models on formalized rules and dynamically retrieve facts for related rules through context-based query generation while delegating decision-making to a symbolic solver.

Speaker Bio:

Prof. Dr. rer. nat. Adrian Paschke is head of the Corporate Semantic Web group with a chair on semantic data intelligence at the institute of computer science, department of mathematics and computer science at Freie Universität Berlin. He additionally is director of the Data Analytics Center at Fraunhofer FOKUS, director of RuleML Inc., and professorial member at the Einstein Center Digital Future, the Dahlem Center for Machine Learning and Robotics, the Institute of Applied Informatics, and founder of the Berlin Semantic Web Meetup group. With over 200 peer-reviewed scientific publications he has made substantial scientific contributions in the field of semantic AI research and is active in standardization of semantic technologies (e.g., RuleML, OASIS LegalRuleML, OMG API4KB, W3C Semantic Web - W3C Rule Interchange Format).