Causal Neurosymbolic AI - A Synergy Between Causal and Neurosymbolic Artificial Intelligence

Watch a PhD defense presentation exploring the integration of causal modeling with neuro-symbolic artificial intelligence to create more human-like reasoning capabilities in AI systems. Discover how the proposed Causal Neuro-Symbolic (Causal NeSy) framework addresses fundamental limitations in current machine learning systems by combining causal Bayesian networks with ontology-encoded structured knowledge to enable machines to understand cause-and-effect relationships and perform counterfactual reasoning. Learn about the framework's architecture that leverages knowledge graphs to create scalable and explainable causal reasoning pipelines, bridging the gap between statistical learning and causal understanding. Examine validation results from the CLEVRER-Humans benchmark dataset and real-world applications in smart manufacturing and autonomous driving that demonstrate enhanced performance and explainability, particularly in limited data scenarios and complex counterfactual situations. Explore how this research advances AI toward more trustworthy, generalizable, and human-aligned intelligent systems capable of meaningful intervention, retrospective explanation, and counterfactual reasoning across high-stakes domains like healthcare, autonomous systems, and manufacturing where causal understanding is essential for decision-making and safety.