KINS: Knowledge Injection via Network Structuring

We propose a novel method to inject symbolic knowledge in form of Datalog formulæ into neural networks (NN), called KINS (Knowledge Injection via Network Structuring). The idea behind our method is to extend NN internal structure with ad-hoc layers built out the injected symbolic knowledge. KINS does not constrain NN to any specific architecture, neither requires logic formulæ to be ground. Moreover, it is robust w.r.t. both lack of data and imperfect/incomplete knowledge. Experiments are reported to demonstrate the potential of KINS.

keywords   neural network, explainable AI, symbolic knowledge injection, KINS, PSyKI

reference talk

 KINS: Knowledge Injection via Network Structuring (CILC 2022, 01/07/2022) — Matteo Magnini (Matteo Magnini, Giovanni Ciatto, Andrea Omicini)

origin event

 CILC 2022

journal or series

 CEUR Workshop Proceedings (CEUR-WS.org)

container publication

 CILC 2022 – Italian Conference on Computational Logic (edited volume, 2022) — Roberta Calegari, Giovanni Ciatto, Andrea Omicini

funding project

 EXPECTATION — Personalized Explainable Artificial Intelligence for decentralized agents with heterogeneous knowledge (01/04/2021–31/03/2024)

reference publication for talk

 KINS: Knowledge Injection via Network Structuring (CILC 2022, 01/07/2022) — Matteo Magnini (Matteo Magnini, Giovanni Ciatto, Andrea Omicini)