An approach to machine learning that uses logic and constructs similar to the way that humans reason through problems. Symbolic learning approaches are not statistically based in the way that deep learning and most of today’s most popular machine learning algorithms work. Rather, symbolic approaches use human-understandable concepts such as names, descriptions, and semantic relationships to reason and deduce meanings. Symbolic approaches to machine learning might be able to handle higher levels of the DIKUW pyramid where statistical approaches to machine learning currently struggle, such as for machine reasoning and common sense. Symbolic approaches to machine learning became popular when Neural Networks and other approaches fell out of favor in the early 1970s. Neural networks and similar statistical approaches to machine learning have regained interest and now symbolic approaches are no longer in favor.
