Siamese recurrent networks learn first-order logic reasoning and exhibit zero-shot compositional generalization
This work addresses the challenge of compositional generalization in neural networks for logic reasoning, showing incremental progress in applying existing methods to a new domain.
The paper tackled the problem of whether neural networks can learn first-order logical entailment by training Siamese Recurrent Networks on an artificial language dataset, achieving near-perfect performance on novel sentences and outperforming recursive networks in recognizing logical relations.
Can neural nets learn logic? We approach this classic question with current methods, and demonstrate that recurrent neural networks can learn to recognize first order logical entailment relations between expressions. We define an artificial language in first-order predicate logic, generate a large dataset of sample 'sentences', and use an automatic theorem prover to infer the relation between random pairs of such sentences. We describe a Siamese neural architecture trained to predict the logical relation, and experiment with recurrent and recursive networks. Siamese Recurrent Networks are surprisingly successful at the entailment recognition task, reaching near perfect performance on novel sentences (consisting of known words), and even outperforming recursive networks. We report a series of experiments to test the ability of the models to perform compositional generalization. In particular, we study how they deal with sentences of unseen length, and sentences containing unseen words. We show that set-ups using LSTMs and GRUs obtain high scores on these tests, demonstrating a form of compositionality.