arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2108.11607cs.CL
收藏

重新思考处理缺失实体注释的负抽样

Rethinking Negative Sampling for Handling Missing Entity Annotations

Tencent AI Lab(腾讯人工智能实验室)

Yangming Li, Lemao Liu, Shuming Shi

摘要:

负采样在处理命名实体识别(NER)的缺失注释方面非常有效。我们的贡献之一是通过引入两个有见地的概念来分析它是如何有意义的:误采样和不确定性。实证研究表明,低误采样率和高的不确定性都是必要的,以实现有前途的性能与负采样。基于命名实体的稀疏性,我们还从理论上推导了零误采样率概率的下界,该下界仅与句子长度有关。另一个贡献是一个自适应和加权的采样分布,通过我们以前的分析,进一步提高了负采样。在合成数据集和注释良好的数据集上的实验(例如,CoNLL-2003)表明,我们提出的方法有利于负采样的F1得分和损失收敛。此外,具有改进的负采样的模型在真实世界数据集上取得了新的最先进的结果(例如,EC)。

英文摘要:

Negative sampling is highly effective in handling missing annotations for named entity recognition (NER). One of our contributions is an analysis on how it makes sense through introducing two insightful concepts: missampling and uncertainty. Empirical studies show low missampling rate and high uncertainty are both essential for achieving promising performances with negative sampling. Based on the sparsity of named entities, we also theoretically derive a lower bound for the probability of zero missampling rate, which is only relevant to sentence length. The other contribution is an adaptive and weighted sampling distribution that further improves negative sampling via our former analysis. Experiments on synthetic datasets and well-annotated datasets (e.g., CoNLL-2003) show that our proposed approach benefits negative sampling in terms of F1 score and loss convergence. Besides, models with improved negative sampling have achieved new state-of-the-art results on real-world datasets (e.g., EC).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑