08/12/2020

Specializing Unsupervised Pretraining Models for Word-Level Semantic Similarity

Anne Lauscher, Ivan Vulić, Edoardo Maria Ponti, Anna Korhonen, Goran Glavaš

Keywords:

Abstract Paper Similar Papers

Abstract: Unsupervised pretraining models have been shown to facilitate a wide range of downstream NLP applications. These models, however, retain some of the limitations of traditional static word embeddings. In particular, they encode only the distributional knowledge available in raw text corpora, incorporated through language modeling objectives. In this work, we complement such distributional knowledge with external lexical knowledge, that is, we integrate the discrete knowledge on word-level semantic similarity into pretraining. To this end, we generalize the standard BERT model to a multi-task learning setting where we couple BERT’s masked language modeling and next sentence prediction objectives with an auxiliary task of binary word relation classification. Our experiments suggest that our “Lexically Informed” BERT (LIBERT), specialized for the word-level semantic similarity, yields better performance than the lexically blind “vanilla” BERT on several language understanding tasks. Concretely, LIBERT outperforms BERT in 9 out of 10 tasks of the GLUE benchmark and is on a par with BERT in the remaining one. Moreover, we show consistent gains on 3 benchmarks for lexical simplification, a task where knowledge about word-level semantic similarity is paramount, as well as large gains on lexical reasoning probes.

The video of this talk cannot be embedded. You can watch it here:
https://underline.io/lecture/6311-specializing-unsupervised-pretraining-models-for-word-level-semantic-similarity
(Link will open in new window)
 0
 0
 0
 0
  Share    
This is an embedded video. Talk and the respective paper are published at COLING 2020 virtual conference. If you are one of the authors of the paper and want to manage your upload, see the question "My papertalk has been externally embedded..." in the FAQ section.

Comments

Post Comment
no comments yet
code of conduct: tbd Characters remaining: 140

Similar Papers

 0
 0
 0
 0
 9:11 
 0
 0
 0
 0
 12:23 
 0
 0
 0
 0
 12:22 
 0
 0
 0
 0
 18:55 
 0
 0
 0
 0
 3:17 
 0
 0
 0
 0
 3:23 
 0
 0
 0
 0
 11:24 
 0
 0
 0
 0
 14:09 
 0
 0
 0
 0
 11:34 
 0
 0
 0
 0
 11:45 
 0
 0
 0
 0
 9:03 
 0
 0
 0
 0
 4:43 
 0
 0
 0
 0
 11:13 
 0
 0
 0
 0
 11:36 
 0
 0
 0
 0
 10:51 
 0
 0
 0
 0
 11:34 
 0
 0
 0
 0
 18:02 
 0
 0
 0
 0
 11:44 
 0
 0
 0
 0
 11:44 
 0
 0
 0
 0
 6:38 
 0
 0
 0
 0
 12:24 
 0
 0
 0
 0
 15:05 
 0
 0
 0
 0
 11:49 
 0
 0
 0
 0
 7:46 
 0
 0
 0
 0
 4:38 
 0
 0
 0
 0
 8:58 
 0
 0
 0
 0
 17:37 
 0
 0
 0
 0
 10:26 
 0
 0
 0
 0
 11:00 
 0
 0
 0
 0
 13:57 
 0
 0
 0
 0
 18:26 
 0
 0
 0
 0
 14:56 
 0
 0
 0
 0
 14:50 
 0
 0
 0
 0
 4:57 
 0
 0
 0
 0
 10:15 
 0
 0
 0
 0
 12:07 
 0
 0
 0
 0
 3:51 
 0
 0
 0
 0
 16:40 
 0
 0
 0
 0
 14:59 
 0
 0
 0
 0
 11:43