Not Far Away, Not So Close: Sample Efficient Nearest Neighbour Data Augmentation via MiniMax

TL;DR

Minimax-kNN is a novel data augmentation method for NLP that efficiently selects the most informative samples to improve knowledge distillation, outperforming existing techniques with fewer examples and less computation.

Contribution

Introduces Minimax-kNN, a dynamic sample selection strategy for kNN-based data augmentation tailored for knowledge distillation in NLP.

Findings

Outperforms strong baselines in text classification tasks.

Requires fewer augmented examples and less computation.

Achieves superior performance over state-of-the-art kNN augmentation.

Abstract

In Natural Language Processing (NLP), finding data augmentation techniques that can produce high-quality human-interpretable examples has always been challenging. Recently, leveraging kNN such that augmented examples are retrieved from large repositories of unlabelled sentences has made a step toward interpretable augmentation. Inspired by this paradigm, we introduce Minimax-kNN, a sample efficient data augmentation strategy tailored for Knowledge Distillation (KD). We exploit a semi-supervised approach based on KD to train a model on augmented data. In contrast to existing kNN augmentation techniques that blindly incorporate all samples, our method dynamically selects a subset of augmented samples that maximizes KL-divergence between the teacher and student models. This step aims to extract the most efficient samples to ensure our augmented data covers regions in the input space with maximum loss value. We evaluated our technique on several text classification tasks and demonstrated that Minimax-kNN consistently outperforms strong baselines. Our results show that Minimax-kNN requires fewer augmented examples and less computation to achieve superior performance over the state-of-the-art kNN-based augmentation techniques.