Zero-Shot Transfer Learning with Synthesized Data for Multi-Domain Dialogue State Tracking
Zero-Shot Transfer Learning with Synthesized Data for Multi-Domain Dialogue State Tracking
复制标题
DOI：
10.18653/v1/2020.acl-main.12
复制
发表时间：
2020-05
期刊：
影响因子：
--
通讯作者：
Giovanni Campagna;Agata Foryciarz;M. Moradshahi;M. Lam
中科院分区：
文献类型：
--
作者：
Giovanni Campagna;Agata Foryciarz;M. Moradshahi;M. Lam
用于多域对话状态跟踪的零触发迁移学习可以让我们处理新的域，而不会产生高成本的数据采集。本文提出了一种新的零-短迁移学习技术，用于对话状态跟踪，其中域内训练数据都是从抽象对话模型和域本体合成的。我们表明，通过合成数据进行数据增强可以提高MultiWOZ 2.1数据集上的TRADE模型和基于BERT的SUMBT模型的零射击学习的准确性。我们发现，仅使用SUMBT模型上的合成域内数据进行训练，可以达到使用完整训练数据集获得的准确度的2/3左右。我们在各个领域的零触发学习平均提高了21%。
Zero-shot transfer learning for multi-domain dialogue state tracking can allow us to handle new domains without incurring the high cost of data acquisition. This paper proposes new zero-short transfer learning technique for dialogue state tracking where the in-domain training data are all synthesized from an abstract dialogue model and the ontology of the domain. We show that data augmentation through synthesized data can improve the accuracy of zero-shot learning for both the TRADE model and the BERT-based SUMBT model on the MultiWOZ 2.1 dataset. We show training with only synthesized in-domain data on the SUMBT model can reach about 2/3 of the accuracy obtained with the full training dataset. We improve the zero-shot learning state of the art on average across domains by 21%.