The Surprising Computational Power of Nondeterministic Stack RNNs
This work addresses the problem of expanding the computational capabilities of neural networks for language modeling and recognition, offering incremental improvements over prior stack-augmented RNNs.
The paper tackles the limited language recognition power of traditional RNNs by augmenting them with a nondeterministic stack, showing that this model can recognize not only context-free languages but also many non-context-free languages and handle larger alphabet sizes than expected, and it proposes a new version with stacks of vectors that improves perplexity on the Penn Treebank benchmark.
Traditional recurrent neural networks (RNNs) have a fixed, finite number of memory cells. In theory (assuming bounded range and precision), this limits their formal language recognition power to regular languages, and in practice, RNNs have been shown to be unable to learn many context-free languages (CFLs). In order to expand the class of languages RNNs recognize, prior work has augmented RNNs with a nondeterministic stack data structure, putting them on par with pushdown automata and increasing their language recognition power to CFLs. Nondeterminism is needed for recognizing all CFLs (not just deterministic CFLs), but in this paper, we show that nondeterminism and the neural controller interact to produce two more unexpected abilities. First, the nondeterministic stack RNN can recognize not only CFLs, but also many non-context-free languages. Second, it can recognize languages with much larger alphabet sizes than one might expect given the size of its stack alphabet. Finally, to increase the information capacity in the stack and allow it to solve more complicated tasks with large alphabet sizes, we propose a new version of the nondeterministic stack that simulates stacks of vectors rather than discrete symbols. We demonstrate perplexity improvements with this new model on the Penn Treebank language modeling benchmark.