Loading…
Glossary term
An approach that combines neural networks with symbolic reasoning.
Neuro-symbolic systems combine neural networks, which are good at perception and pattern recognition, with symbolic reasoning, which is good at logic, rules and exact calculation. For example, a network reads an image and a logic engine answers a question about it. The aim is systems that are both flexible and reliable.
It is a leading idea for making AI more robust, explainable and data-efficient.