Synthesizing Context-free Grammars from Recurrent Neural Networks (Extended Version)
This work addresses the need for understanding and verifying RNNs in natural language processing, though it is incremental as it builds on existing automata extraction techniques.
The authors tackled the problem of extracting interpretable context-free grammars from trained recurrent neural networks, resulting in a method that converts RNNs into CFGs to improve classification accuracy as recursion depth increases.
We present an algorithm for extracting a subclass of the context free grammars (CFGs) from a trained recurrent neural network (RNN). We develop a new framework, pattern rule sets (PRSs), which describe sequences of deterministic finite automata (DFAs) that approximate a non-regular language. We present an algorithm for recovering the PRS behind a sequence of such automata, and apply it to the sequences of automata extracted from trained RNNs using the L* algorithm. We then show how the PRS may converted into a CFG, enabling a familiar and useful presentation of the learned language. Extracting the learned language of an RNN is important to facilitate understanding of the RNN and to verify its correctness. Furthermore, the extracted CFG can augment the RNN in classifying correct sentences, as the RNN's predictive accuracy decreases when the recursion depth and distance between matching delimiters of its input sequences increases.