Learning Explanations From Language Data

David Harbecke, Robert Schwarzenberg, Christoph Alt

In: EMNLP Workshop on Interpreting and Analysing Neural Networks for NLP (BlackboxNLP). Conference on Emperical Methods in Natural Language Processing (EMNLP-2018) October 31-November 4 Brussels Belgium EMNLP 2018.


PatternAttribution is a recent method, introduced in the vision domain, that explains classifications of deep neural networks. We demonstrate that it also generates meaningful interpretations in the language domain.


emnlp-learning-explanations.pdf (pdf, 162 KB)

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz