Do Neural Nets Learn Statistical Laws behind Natural Language?
This provides empirical insights into the capabilities and limitations of neural networks for language processing, potentially guiding architectural improvements.
The paper investigated whether neural language models learn fundamental statistical properties of natural language, finding that an LSTM-based model effectively reproduces Zipf's law and Heaps' law but fails to capture long-range correlations.
The performance of deep learning in natural language processing has been spectacular, but the reasons for this success remain unclear because of the inherent complexity of deep learning. This paper provides empirical evidence of its effectiveness and of a limitation of neural networks for language engineering. Precisely, we demonstrate that a neural language model based on long short-term memory (LSTM) effectively reproduces Zipf's law and Heaps' law, two representative statistical properties underlying natural language. We discuss the quality of reproducibility and the emergence of Zipf's law and Heaps' law as training progresses. We also point out that the neural language model has a limitation in reproducing long-range correlation, another statistical property of natural language. This understanding could provide a direction for improving the architectures of neural networks.