Learning to Augment for Data-Scarce Domain BERT Knowledge Distillation

TL;DR

This paper introduces a novel data augmentation approach for domain-specific BERT knowledge distillation, improving student model performance in data-scarce scenarios by leveraging source domain data and reinforcement learning.

Contribution

It proposes a cross-domain augmentation method with a reinforced selector to enhance knowledge transfer in data-scarce domain distillation tasks.

Findings

Significantly outperforms state-of-the-art baselines on four tasks.

Student models outperform large teachers with fewer parameters.

Effective in scenarios with limited labeled data.

Abstract

Despite pre-trained language models such as BERT have achieved appealing performance in a wide range of natural language processing tasks, they are computationally expensive to be deployed in real-time applications. A typical method is to adopt knowledge distillation to compress these large pre-trained models (teacher models) to small student models. However, for a target domain with scarce training data, the teacher can hardly pass useful knowledge to the student, which yields performance degradation for the student models. To tackle this problem, we propose a method to learn to augment for data-scarce domain BERT knowledge distillation, by learning a cross-domain manipulation scheme that automatically augments the target with the help of resource-rich source domains. Specifically, the proposed method generates samples acquired from a stationary distribution near the target data and adopts a reinforced selector to automatically refine the augmentation strategy according to the performance of the student. Extensive experiments demonstrate that the proposed method significantly outperforms state-of-the-art baselines on four different tasks, and for the data-scarce domains, the compressed student models even perform better than the original large teacher model, with much fewer parameters (only ${\sim}13.3\%$) when only a few labeled examples available.