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


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(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: Improving Contrastive Learning of Sentence Embeddings with Case-Augmented Positives and Retrieved Negatives

Abstract: Following SimCSE, contrastive learning based methods have achieved the state-of-the-art (SOTA) performance in learning sentence embeddings. However, the unsupervised contrastive learning methods still lag far behind the supervised counterparts. We attribute this to the quality of positive and negative samples, and aim to improve both. Specifically, for positive samples, we propose switch-case augmentation to flip the case of the first letter of randomly selected words in a sentence. This is to counteract the intrinsic bias of pre-trained token embeddings to frequency, word cases and subwords. For negative samples, we sample hard negatives from the whole dataset based on a pre-trained language model. Combining the above two methods with SimCSE, our proposed Contrastive learning with Augmented and Retrieved Data for Sentence embedding (CARDS) method significantly surpasses the current SOTA on STS benchmarks in the unsupervised setting.
Comments: 7 pages, 3 figures, 6 tables. Accepted to SIGIR 22. Code at this https URL
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
ACM classes: H.3.3
DOI: 10.1145/3477495.3531823
Cite as: arXiv:2206.02457 [cs.CL]
  (or arXiv:2206.02457v1 [cs.CL] for this version)

Submission history

From: Wei Wang [view email]
[v1] Mon, 6 Jun 2022 09:46:12 GMT (760kb,D)

Link back to: arXiv, form interface, contact.