LST: Lexicon-Guided Self-Training for Few-Shot Text Classification
This work addresses the challenge of improving self-training for few-shot text classification, offering a domain-specific incremental advance.
The paper tackles the problem of overconfident pseudo-labeling in few-shot text classification by introducing LST, a lexicon-guided self-training method that refines the lexicon using confidence predictions, achieving 1.0-2.0% better performance on five benchmark datasets with 30 labeled samples per class.
Self-training provides an effective means of using an extremely small amount of labeled data to create pseudo-labels for unlabeled data. Many state-of-the-art self-training approaches hinge on different regularization methods to prevent overfitting and improve generalization. Yet they still rely heavily on predictions initially trained with the limited labeled data as pseudo-labels and are likely to put overconfident label belief on erroneous classes depending on the first prediction. To tackle this issue in text classification, we introduce LST, a simple self-training method that uses a lexicon to guide the pseudo-labeling mechanism in a linguistically-enriched manner. We consistently refine the lexicon by predicting confidence of the unseen data to teach pseudo-labels better in the training iterations. We demonstrate that this simple yet well-crafted lexical knowledge achieves 1.0-2.0% better performance on 30 labeled samples per class for five benchmark datasets than the current state-of-the-art approaches.