We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CL

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Interpretation of NLP models through input marginalization

Abstract: To demystify the "black box" property of deep neural networks for natural language processing (NLP), several methods have been proposed to interpret their predictions by measuring the change in prediction probability after erasing each token of an input. Since existing methods replace each token with a predefined value (i.e., zero), the resulting sentence lies out of the training data distribution, yielding misleading interpretations. In this study, we raise the out-of-distribution problem induced by the existing interpretation methods and present a remedy; we propose to marginalize each token out. We interpret various NLP models trained for sentiment analysis and natural language inference using the proposed method.
Comments: 10 pages, 5 figures, to be published in the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP 2020)
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Cite as: arXiv:2010.13984 [cs.CL]
  (or arXiv:2010.13984v1 [cs.CL] for this version)

Submission history

From: Siwon Kim [view email]
[v1] Tue, 27 Oct 2020 01:40:41 GMT (2088kb,D)

Link back to: arXiv, form interface, contact.