The Hardest Challenge in Neurosymbolic AI: Symbol Grounding

Abstract

While ML models can provide excellent results from a statistical sense, their errors cause issues when model output is used as input into a symbolic system. For example, a small error in getting a handwritten digit wrong can lead to catastrophic failure. The key issue is converting perceptual input into symbols part of a “symbol system”. Researchers have studied this problem since the 1990’s and it has become more relevant in AI with the increased popularity of neurosymbolic approaches.

About the channel:

The Neuro Symbolic Channel provides the tutorials, courses, and research results on one of the most exciting areas in artificial intelligence and machine learning. With content originally from the AI course taught at Arizona State University, this channel brings you the latest at the intersection of symbolic methods (e.g., logic programming) and deep learning. Learn about the latest algorithms, Python packages, and progress toward larger goals such as artificial general intelligence (AGI).