Finding Syntactic Representations in Neural Stacks
This addresses the interpretability challenge in neural networks for linguists and AI researchers, but it is incremental as it builds on existing techniques to validate known biases.
The paper tackled the problem of assessing whether neural networks with differentiable stacks learn hierarchical structure by extracting syntactic trees from stack RNNs trained on language tasks, finding that the models produce parses reflecting natural language syntax.
Neural network architectures have been augmented with differentiable stacks in order to introduce a bias toward learning hierarchy-sensitive regularities. It has, however, proven difficult to assess the degree to which such a bias is effective, as the operation of the differentiable stack is not always interpretable. In this paper, we attempt to detect the presence of latent representations of hierarchical structure through an exploration of the unsupervised learning of constituency structure. Using a technique due to Shen et al. (2018a,b), we extract syntactic trees from the pushing behavior of stack RNNs trained on language modeling and classification objectives. We find that our models produce parses that reflect natural language syntactic constituencies, demonstrating that stack RNNs do indeed infer linguistically relevant hierarchical structure.