Graph Neural Networks and Boolean Satisfiability

In this paper we explore whether or not deep neural architectures can learn to classify Boolean satisfiability (SAT). We devote considerable time to discussing the theoretical properties of SAT. Then, we define a graph representation for Boolean formulas in conjunctive normal form, and train neural classifiers over general graph structures called Graph Neural Networks, or GNNs, to recognize features of satisfiability. To the best of our knowledge this has never been tried before. Our preliminary findings are potentially profound. In a weakly-supervised setting, that is, without problem specific feature engineering, Graph Neural Networks can learn features of satisfiability.

[1]  Toby Walsh,et al.  Restart Strategy Selection Using Machine Learning Techniques , 2009, SAT.

[2]  Andrew Y. Ng,et al.  Semantic Compositionality through Recursive Matrix-Vector Spaces , 2012, EMNLP.

[3]  Stephen A. Cook,et al.  The complexity of theorem-proving procedures , 1971, STOC.

[4]  Ah Chung Tsoi,et al.  Computational Capabilities of Graph Neural Networks , 2009, IEEE Transactions on Neural Networks.

[5]  Peter Kulchyski.  and , 2015 .

[6]  J. J. Hopfield,et al.  “Neural” computation of decisions in optimization problems , 1985, Biological Cybernetics.

[7]  Ah Chung Tsoi,et al.  The Graph Neural Network Model , 2009, IEEE Transactions on Neural Networks.

[8]  Michèle Sebag,et al.  Phase Transitions in Machine Learning , 2017, Encyclopedia of Machine Learning and Data Mining.

[9]  Barry O’Sullivan,et al.  Satisfiability as a Classification Problem ? , 2008 .

[10]  Bart Selman,et al.  Local search strategies for satisfiability testing , 1993, Cliques, Coloring, and Satisfiability.

[11]  James L. Johnson.  A Neural Network Approach to the 3-Satisfiability Problem , 1989, J. Parallel Distributed Comput..