arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2112.09427eess.AScs.CLcs.LGstat.ML
收藏

连续学习用于单语端到端自动语音识别

Continual Learning for Monolingual End-to-End Automatic Speech Recognition

KU Leuven(鲁汶大学)

Steven Vander Eeckt, Hugo Van hamme

AI总结:

本文提出了一种连续学习方法,用于在不存储历史数据的情况下提升单语端到端ASR模型在多个新任务上的性能。

摘要:

将自动语音识别(ASR)模型适应到新领域会导致在原始领域上的性能下降,这种现象称为灾难性遗忘(CF)。即使是单语ASR模型,在没有遭受CF的情况下也无法扩展到新的口音、方言、主题等,这使得它们无法在不存储所有先前数据的情况下持续改进。幸运的是,连续学习(CL)方法旨在实现持续适应并克服CF,可以被用来。在本文中,我们实现了大量的CL方法用于端到端ASR,并测试和比较它们在四个新任务上扩展单语混合CTC-Transformer模型的能力。我们发现,表现最好的CL方法能够将细调模型(下限)与在所有任务上联合训练的模型(上限)之间的差距缩小超过40%,同时仅需要访问原始数据的0.6%。

英文摘要:

Adapting Automatic Speech Recognition (ASR) models to new domains results in a deterioration of performance on the original domain(s), a phenomenon called Catastrophic Forgetting (CF). Even monolingual ASR models cannot be extended to new accents, dialects, topics, etc. without suffering from CF, making them unable to be continually enhanced without storing all past data. Fortunately, Continual Learning (CL) methods, which aim to enable continual adaptation while overcoming CF, can be used. In this paper, we implement an extensive number of CL methods for End-to-End ASR and test and compare their ability to extend a monolingual Hybrid CTC-Transformer model across four new tasks. We find that the best performing CL method closes the gap between the fine-tuned model (lower bound) and the model trained jointly on all tasks (upper bound) by more than 40%, while requiring access to only 0.6% of the original data.

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。

↑