BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//TYPO3/NONSGML Calendarize//EN


BEGIN:VEVENT
UID:calendarize-guest-seminar-neuro-symbolic-methods-for-semantic-web-ontologies-1
DTSTAMP:20260512T134338Z

DTSTART:20260515T070000Z
DTEND:20260515T090000Z
    
SUMMARY:Guest Seminar “Neuro-symbolic methods for Semantic Web ontologies”
DESCRIPTION:The speaker will be Robert Hoehndorf\, Associate Professor of Computer Science at King Abdullah University of Science and Technology\, where he leads the Bio-Ontology Research Group (BORG). His research focuses on neuro-symbolic AI\, combining formal knowledge representation with machine learning for applications in functional genomics\, biomedical informatics\, and causal inference. He also serves as Editor-in-Chief of the Journal of Biomedical Semantics.\nIn his talk\, he will explore how Semantic Web ontologies are used to provide conceptual frameworks for sharing and integrating data through logic-based languages. These ontologies can also serve as background knowledge in machine learning models or define domain-specific constraints that support automated verification and zero-shot predictions.\nRobert Hoehndorf explains that neuro-symbolic AI systems rely on two key components: embedding symbolic representations such as ontologies into machine learning models\, and extracting symbolic knowledge back from these embeddings. He will introduce methods for embedding Semantic Web ontologies and discuss how their properties relate to model theory and proof theory. Furthermore\, he will demonstrate how embeddings can be inverted to extract axioms and enable approximate reasoning.\nThe seminar will also highlight practical applications in the life sciences\, where large ontologies are widely used. In particular\, he will present how neuro-symbolic methods can be applied to predict protein functions using the Gene Ontology.
X-ALT-DESC;FMTTYPE=text/html:<p>The speaker will be Robert Hoehndorf\, Associate Professor of Computer Science at King Abdullah University of Science and Technology\, where he leads the Bio-Ontology Research Group (BORG). His research focuses on neuro-symbolic AI\, combining formal knowledge representation with machine learning for applications in functional genomics\, biomedical informatics\, and causal inference. He also serves as Editor-in-Chief of the Journal of Biomedical Semantics.</p>\n<p>In his talk\, he will explore how Semantic Web ontologies are used to provide conceptual frameworks for sharing and integrating data through logic-based languages. These ontologies can also serve as background knowledge in machine learning models or define domain-specific constraints that support automated verification and zero-shot predictions.</p>\n<p>Robert Hoehndorf explains that neuro-symbolic AI systems rely on two key components: embedding symbolic representations such as ontologies into machine learning models\, and extracting symbolic knowledge back from these embeddings. He will introduce methods for embedding Semantic Web ontologies and discuss how their properties relate to model theory and proof theory. Furthermore\, he will demonstrate how embeddings can be inverted to extract axioms and enable approximate reasoning.</p>\n<p>The seminar will also highlight practical applications in the life sciences\, where large ontologies are widely used. In particular\, he will present how neuro-symbolic methods can be applied to predict protein functions using the Gene Ontology.</p>
LOCATION:University of Latvia House of Science\, Room 501
END:VEVENT

END:VCALENDAR