neurosymbolic AI

Terms from Artificial Intelligence: humans at the heart of algorithms

Neurosymbolic AI refers to techniques that combine sub-symbolic neural networks with symbolic techniques. These may be fairly distinct modules in a hybrid archtecture, but also can be integrated at.a more fundamental way such as using neural techniques to reason over semantic networks or the way high-level counterfactual reasonng is used to guide low-level stimulus-response learning in the cognitive model of regret.

Used in Chap. 6: page 85; Chap. 24: page 376

Also known as neurosymbolic