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

Computer Science > Computation and Language

Title: Aligning Faithful Interpretations with their Social Attribution

Abstract: We find that the requirement of model interpretations to be faithful is vague and incomplete. With interpretation by textual highlights as a case-study, we present several failure cases. Borrowing concepts from social science, we identify that the problem is a misalignment between the causal chain of decisions (causal attribution) and the attribution of human behavior to the interpretation (social attribution). We re-formulate faithfulness as an accurate attribution of causality to the model, and introduce the concept of aligned faithfulness: faithful causal chains that are aligned with their expected social behavior. The two steps of causal attribution and social attribution together complete the process of explaining behavior. With this formalization, we characterize various failures of misaligned faithful highlight interpretations, and propose an alternative causal chain to remedy the issues. Finally, we implement highlight explanations of the proposed causal format using contrastive explanations.
Comments: Accepted as a journal paper to TACL
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Cite as: arXiv:2006.01067 [cs.CL]
  (or arXiv:2006.01067v3 [cs.CL] for this version)

Submission history

From: Alon Jacovi [view email]
[v1] Mon, 1 Jun 2020 16:45:38 GMT (508kb,D)
[v2] Sun, 3 Jan 2021 00:45:44 GMT (524kb,D)
[v3] Thu, 14 Jan 2021 18:54:01 GMT (524kb,D)

Link back to: arXiv, form interface, contact.