•    Knowledge-Infused Learning: Strategies for using knowledge with Deep Learning and NLP, with some applications


       A talk by Professor Amit Sheth
       [Google Drive Link for the talk]



    The symbolic AI approach was overtaken by the statistical AI approach in this century. The deep learning models utilizing massive computing power and enormous datasets have significantly outperformed prior historical benchmarks on increasingly difficult, well-defined research tasks across technology domains such as computer vision, natural language processing, signal processing, and human-computer interactions. Following this success, the last few years have seen a growing interest in combining knowledge with machine learning (hence combine symbolic and statistical approaches) for several reasons, including (a) reducing the need for labeling, (b) gaining broader coverage through the use of knowledge in the forms of existing ontologies or knowledge graphs, (c) further improving the performance on both qualitative or functional and quantitative measures, and (d) improve the system’s interpretability and explainability. Although not the first use of knowledge graph for semantic search, Google demonstrated the importance of knowledge graphs for functional gains over systems that primarily relied on machine learning. In the last decade, more structured knowledge have been repurposed as knowledge graphs, and we have significantly improved our ability to create and maintain knowledge graphs that capture domain knowledge and human experiences.

    In this talk, I describe knowledge-infused learning - a variety of shallow, semi-deep, and deep approaches to infuse or combine knowledge with deep learning. I will briefly discuss how this makes a fundamental difference in the interpretability and explainability of current approaches and illustrate a few applications do demonstrate the benefits of the emerging hybrid or neuro-symbolic AI approaches that exploit knowledge graphs and deep learning. For more details, visit here.

    Index Terms: Knowledge Graphs, Knowledge Infusion, NeuroSymbolic AI, Explainability, Interpretability, Black-Box Deep Learning

    Bio

    Prof. Amit Sheth is an Educator, Researcher, and Entrepreneur. Before he joined the University of South Carolina as the founding director of the university-wide AI Institute, he was the LexisNexis Ohio Eminent Scholar and executive director of Ohio Center of Excellence in Knowledge-enabled Computing at the Wright State University during 2007 and 2019. He is a Fellow of IEEE, AAAI, and AAAS. He was awarded IEEE TCSVC’s Research Innovation Award for pioneering and enduring research, applications, and adoption of distributed workflow processes and semantics in services computing. He is among the highly cited computer scientists worldwide. He has (co-)founded four companies, three of them by licensing his university research outcomes, including the first Semantic Web company in 1999 that pioneered technology similar to what is found today in Google Semantic Search and Knowledge Graph. He is particularly proud of the success of his 30 former Ph.D. advisees and >10 postdocs in academia, industry research, and entrepreneurs.