Achille Fokoue

Achille Fokoue is a Distinguished Research Scientist, Master Inventor, and Research Manager at IBM Research AI in Yorktown Heights, NY, where he leads the Foundations of AI Reasoning group. He has over 19 years of research experience in knowledge representation and reasoning focused on developing theories, algorithms, standards, and systems for scaling reasoning over large and expressive knowledge bases that tolerate inconsistencies and uncertainties inherent in knowledge bases populated from unstructured sources. He has led various research efforts on applying machine learning and knowledge representation and reasoning in many domains. He is a co-editor of the OWL 2 Web Ontology Language Profiles specification and has authored or co-authored over 100+ scientific reports and manuscripts that have been cited, in aggregate, more than 3600 times.

ABSTRACT

Answering Questions using Neuro-Symbolic AI (in an explainable way)

Organizations and individuals alike are faced with an increased volume, velocity, and variety of their data. They often struggle to answer basic questions about their data and gain insight from it. Given enough training data and supervision, traditional deep learning can help address this problem. Unfortunately, large training data can be very expensive in many domains, and many stakeholders lack the deep technical skills to interact and query the data using structured query languages. A key question is then how can we gain insight and answer questions over the growing volume of complex data with limited technical expertise and limited supervision? In this talk, I will argue that a Neuro-Symbolic AI approach that combines traditional deep learning and symbolic reasoning can learn, with limited supervision, how to answer questions about what is explicitly in the data, what is implied by the data, and what is missing from the data.