Bridging symbolic and subsymbolic reasoning with minimax entropy models

Article type: Research Article

Authors: Marra, Giuseppe^{; *}

Affiliations: Computer Science Department, KU Leuven, Celestijnenlaan 200A, Leuven, Belgium

Correspondence: [*] Corresponding author: Giuseppe Marra, Computer Science Department, KU Leuven, Celestijnenlaan 200A, 3001 Leuven, Belgium. E-mail: [email protected].

Abstract: In this paper, we investigate MiniMax Entropy models, a class of neural symbolic models where symbolic and subsymbolic features are seamlessly integrated. We show how these models recover classical algorithms from both the deep learning and statistical relational learning scenarios. Novel hybrid settings are defined and experimentally explored, showing state-of-the-art performance in collective classification, knowledge base completion and graph (molecular) data generation.

Keywords: Neural symbolic artificial intelligence, statistical relational artificial intelligence, deep learning, logic

DOI: 10.3233/IA-210088

Journal: Intelligenza Artificiale, vol. 15, no. 2, pp. 71-90, 2021