Neuro-Symbolic Artificial Intelligence: A Task-Directed Survey in the Black-Box Models Era

This academic survey examines Neuro-Symbolic AI methods that combine neural networks with symbolic computing to enhance explainability and reasoning capabilities. The research explores how these hybrid approaches can address limitations in semantic generalizability and compete with pure connectionist systems in real-world applications.