Neuro-Symbolic AI, which is alternatively called composite AI, is a relatively new term for a well-established concept with enormous significance for almost any enterprise application of Artificial Intelligence. By combining AI’s statistical foundation (exemplified by machine learning) with its knowledge foundation (exemplified by knowledge graphs and rules), organizations get the most effective cognitive analytics results with the least amount of headaches—and cost.

Pairing these two historical pillars of AI is essential to maximizing investments in these technologies and in data themselves. Alone, machine learning simply patterns recognition at a massive scale.

By itself, rules-based symbolic reasoning doesn’t improve over time. Together, these AI approaches create total machine intelligence with logic-based systems that get better with each application.

Understanding Neuro-Symbolic AI

The full value of Neuro-Symbolic AI isn’t just in its elimination of the training data or taxonomy building delays that otherwise impede Natural Language Processing applications, cognitive search, or conversational AI. Nor is it only in the ease of generating queries and bettering the results of constraint systems, all of which it inherently does. The real reason for the adoption of composite AI is that, as Marvin Minsky alluded to in his society of mind metaphor, human intelligence is comprised of numerous systems (analogous to diverse society members or machines) working together to produce intelligent behavior. Similarly, AI requires an assortment of approaches and techniques working in conjunction to solve the myriad business problems organizations regularly apply to it.

Natural Language Proficiency

Knowledge graphs are seminal to Neuro-Symbolic AI because they represent enterprise concepts via data so intelligent systems can reason and learn about them. These graphs are the foundation of the unified system AI pioneer Allen Newell who claims that the layers of mechanisms are required for intelligence. Nearly all the various AI elements—including semantic inferencing, unsupervised learning, supervised learning, and other reasoning and statistical approaches—are readily incorporated or visualized in knowledge graphs containing business logic for enterprise data.

These semantic graphs are integral to the most widely used Neuro-Symbolic AI application: natural language question-answering of textual data (including transcriptions of speech).

Combining AI’s reasoning and learning sides generates profound results in this deployment. It allows users to rapidly construct a knowledge base from a corpus via unsupervised and supervised learning techniques, the former of which drastically reduce the amounts of labeled training data impeding pure supervised learning approaches. Alternatively, combining AI’s statistical and knowledge sides can automate the labels for supervised learning approaches to natural language applications, which also saves money and time. No matter which technique is used, the marketplace is realizing that natural language applications almost always need some reasoning and knowledge-based techniques to deliver the best outcomes at the lowest costs.

Machine Logic

Knowledge graphs are also central to Neuro-Symbolic AI because they provide ideal settings for machine logic. Their heightened relationship detection and intelligent inferences make them complementary for logic-based systems like Prolog, an AI language specializing in first-order logic. Consequently, organizations can write various AI algorithms in this language that’s also useful for creating logic rules, which have a lengthy history in AI via symbolic reasoning. Newell proposed that human cognition could be expressed in a system of symbols that could provide rules-based constraints.

Knowledge graphs can do the same thing for AI, allowing organizations to create rules about different ways to identify a business outcome (like codifying rules for finding out if a customer mentioned a specific timeframe for making a purchase to contact center representatives). This capability is also applicable to constraint system deployments like optimizing delivery routes and the time they take. These systems’ results get better by running machine learning on the outcomes, then inputting the latter findings into the graph as additional knowledge factoring into rules.

Graph Neural Networks

Another compelling use case for composite AI is the predictive accuracy of Graph Neural Networks (GNNs). These machine learning models excel in graph settings that fully depict or visualize all of the dimensions of their statistical analytics on multi-dimensionality data—like considerations across networks for fraud analysis, for example.

More importantly, knowledge graph environments allow organizations to combine traditional knowledge-based techniques like semantic inferencing with the node prediction prowess of GNNs to ascertain the likelihood of customers buying a certain product at a specific time, for example. This combination not only simplifies the query writing process for analyzing customer subsets or micro-segments, but also grants unparalleled insight into graph influencers and how they’ll affect business use cases.

Composite AI

Organizations’ adoption of Neuro-Symbolic AI as a product is still growing. However, the overall utility—and use cases—from the concept of coupling AI’s statistical and knowledge base is too great to ignore and one of the most effective means of creating competitive advantage that these technologies afford.

[To share your insights with us, please write to sghosh@martechseries.com]