Explaining Character-Aware Neural Networks for Word-Level Prediction: Do They Discover Linguistic Rules?

By Fréderic Godin, Kris Demuynck, Joni Dambre, Wesley De Neve, Thomas Demeester. 2018

18 citations recorded by the source corpus.

Character-level features are currently used in different neural network-based natural language processing algorithms. However, little is known about the character-level patterns those models learn. Moreover, models are often compared only quantitatively while a qualitative analysis is missing. In this paper, we investigate which character-level patterns neural networks learn and if those patterns coincide with manually-defined word segmentations and annotations. To that end, we extend the contextual decomposition technique (Murdoch et al. 2018) to convolutional neural networks which allows us to compare convolutional neural networks and bidirectional long short-term memory networks. We evaluate and compare these models for the task of morphological tagging on three morphologically different languages and show that these models implicitly discover understandable linguistic rules.

Subfields: Artificial Intelligence, Natural Language Processing, Safety, Robustness & Interpretability, Computational Linguistics & Semantics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In Safety, Robustness & Interpretability

In Computational Linguistics & Semantics

Citations in the corpus

Analysis Methods in Neural Language Processing: A Survey (2018, background): “The important or salient features can then be visualized in selected examples (Li et al., 2016a; Aubakirova and Bansal, 2016; Sundararajan et al., 2017; Arras et al., 2017a,b; Ding et al., 2017; Murdoch et al., 2018; Mudrakarta et al., 2018; Montavon et al., 2018; Godin et al., 2018).”
A Survey of the State of Explainable AI for Natural Language Processing (2020, background): “…can be built on different types of features, such as manual features obtained from feature engineering (e.g., Voskarides et al., 2015), lexical features including word/tokens and n-gram (e.g., Godin et al., 2018; Mullenbach et al., 2018), or latent features learned by NNs (e.g., Xie et al., 2017).”
Hierarchical interpretations for neural network predictions (2018, methodology): “25 Accepted as a conference paper at ICLR 2019 Figure S5: Comparing unit-level CD scores to CD scores from the naive extension of CD to CNNs, independently developed by Godin et al. (2018).”