Pascal Hitzler: Towards Explaining Neural Networks Through Background Knowledge

23. Mai 2017
CITEC, 2.015

Towards Explaining Neural Networks Through Background Knowledge

Pascal Hitzler

The ever increasing prevalence of publicly available structured data on the World Wide Web enables new applications in a variety of domains. In this presentation, we provide a conceptual approach that leverages such data in order to explain the input-output behavior of trained artificial neural networks. We apply existing Semantic Web technologies in order to provide an experimental proof of concept. The presentation starts by investigating the case of propositional rule extraction as a base case, carrying past results over to incorporate background knowledge. In the second part, we incorporate knowledge graphs and ontologies and show how the DL-Learner symbolic machine learning system can be used to generate explanations which take such background knowledge into account.

Presenter's Bio:

Pascal Hitzler is endowed NCR Distinguished Professor and Director of Data Science at the Department of Computer Science and Engineering at Wright State University in Dayton, Ohio, U.S.A. His research record lists over 350 publications in such diverse areas as semantic web, neural-symbolic integration, knowledge representation and reasoning, machine learning, denotational semantics, and set-theoretic topology. He is Editor-in-chief of the Semantic Web journal by IOS Press, and of the IOS Press book series Studies on the Semantic Web. He is co-author of the W3C Recommendation OWL 2 Primer, and of the book Foundations of Semantic Web Technologies by CRC Press, 2010 which was named as one out of seven Outstanding Academic Titles 2010 in Information and Computer Science by the American Library Association's Choice Magazine, and has translations into German and Chinese. He is on the editorial board of several journals and book series and is a founding steering committee member of the Neural-Symbolic Learning and Reasoning (NeSy) workshop series, and of the Association for Ontology Design and Patterns (ODPA). He also frequently acts as conference chair in various functions. For more information, see