Embedding Hallucination for Few-Shot Language Fine-tuning

TL;DR

EmbedHalluc is a novel data augmentation technique for few-shot language fine-tuning that generates auxiliary embeddings to improve model generalization and reduce over-fitting, outperforming existing methods.

Contribution

We introduce EmbedHalluc, a method that creates synthetic embedding-label pairs via adversarial training to enhance few-shot language model fine-tuning.

Findings

EmbedHalluc improves performance across various language tasks.

It outperforms data augmentation, pseudo-labeling, and regularization methods.

The approach effectively mitigates over-fitting in few-shot learning.

Abstract

Few-shot language learners adapt knowledge from a pre-trained model to recognize novel classes from a few-labeled sentences. In such settings, fine-tuning a pre-trained language model can cause severe over-fitting. In this paper, we propose an Embedding Hallucination (EmbedHalluc) method, which generates auxiliary embedding-label pairs to expand the fine-tuning dataset. The hallucinator is trained by playing an adversarial game with the discriminator, such that the hallucinated embedding is indiscriminative to the real ones in the fine-tuning dataset. By training with the extended dataset, the language learner effectively learns from the diverse hallucinated embeddings to overcome the over-fitting issue. Experiments demonstrate that our proposed method is effective in a wide range of language tasks, outperforming current fine-tuning methods. Further, we show that EmbedHalluc outperforms other methods that address this over-fitting problem, such as common data augmentation, semi-supervised pseudo-labeling, and regularization. The code will be made available at: https://github.com/yiren-jian/EmbedHalluc.