Learning from Multiple Noisy Augmented Data Sets for Better Cross-Lingual Spoken Language Understanding
This work addresses data scarcity and noise issues in scaling spoken language understanding to low-resource languages, representing an incremental advancement in data augmentation techniques.
The paper tackles the problem of noisy augmented data in cross-lingual spoken language understanding for low-resource languages by developing a denoising training approach, resulting in performance improvements of 3.05 and 4.24 percentage points over state-of-the-art methods on two benchmark datasets.
Lack of training data presents a grand challenge to scaling out spoken language understanding (SLU) to low-resource languages. Although various data augmentation approaches have been proposed to synthesize training data in low-resource target languages, the augmented data sets are often noisy, and thus impede the performance of SLU models. In this paper we focus on mitigating noise in augmented data. We develop a denoising training approach. Multiple models are trained with data produced by various augmented methods. Those models provide supervision signals to each other. The experimental results show that our method outperforms the existing state of the art by 3.05 and 4.24 percentage points on two benchmark datasets, respectively. The code will be made open sourced on github.