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

Download:

Current browse context:

cs.CL

Change to browse by:

cs

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: Relation Extraction with Weighted Contrastive Pre-training on Distant Supervision

Abstract: Contrastive pre-training on distant supervision has shown remarkable effectiveness for improving supervised relation extraction tasks. However, the existing methods ignore the intrinsic noise of distant supervision during the pre-training stage. In this paper, we propose a weighted contrastive learning method by leveraging the supervised data to estimate the reliability of pre-training instances and explicitly reduce the effect of noise. Experimental results on three supervised datasets demonstrate the advantages of our proposed weighted contrastive learning approach, compared to two state-of-the-art non-weighted baselines.
Comments: Under review
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2205.08770 [cs.CL]
  (or arXiv:2205.08770v1 [cs.CL] for this version)

Submission history

From: Zhen Wan [view email]
[v1] Wed, 18 May 2022 07:45:59 GMT (212kb,D)

Link back to: arXiv, form interface, contact.