TSDAE: Using Transformer-based Sequential Denoising Auto-Encoder for Unsupervised Sentence Embedding Learning

14 Apr 2021  ·  Kexin Wang, Nils Reimers, Iryna Gurevych ·

Learning sentence embeddings often requires a large amount of labeled data. However, for most tasks and domains, labeled data is seldom available and creating it is expensive. In this work, we present a new state-of-the-art unsupervised method based on pre-trained Transformers and Sequential Denoising Auto-Encoder (TSDAE) which outperforms previous approaches by up to 6.4 points. It can achieve up to 93.1% of the performance of in-domain supervised approaches. Further, we show that TSDAE is a strong domain adaptation and pre-training method for sentence embeddings, significantly outperforming other approaches like Masked Language Model. A crucial shortcoming of previous studies is the narrow evaluation: Most work mainly evaluates on the single task of Semantic Textual Similarity (STS), which does not require any domain knowledge. It is unclear if these proposed methods generalize to other domains and tasks. We fill this gap and evaluate TSDAE and other recent approaches on four different datasets from heterogeneous domains.

PDF Abstract

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Re-Ranking AskUbuntu TSDAE MAP 59.4 # 1
Information Retrieval CQADupStack TSDAE mAP@100 0.145 # 3
Paraphrase Identification PIT TSDAE AP 69.2 # 1
Re-Ranking SciDocs TSDAE Cite 71.4 # 1
CC 73.9 # 1
CR 75.0 # 1
CV 75.6 # 1
Paraphrase Identification TURL TSDAE AP 76.8 # 1