一种基于元学习的医疗文本分类模型

医疗文本专业术语复杂，垂直领域训练样本不足，传统的分类方法不能满足现实需求，提出一种基于元学习的小样本文本分类模型提高医疗文本分类效率。该模型基于迁移学习思想，加入注意力机制赋予句子中的词语不同的权重，利用两个相互竞争的神经网络分别扮演领域识别者和元知识生成者的角色，通过自适应性网络加强元学习对新数据集的适应性，最后使用岭回归获得数据集的分类。实验对比分析结果验证了该模型对一些公开文本数据集和医疗文本数据具有很好的分类效果。基于元学习的小样本文本分类模型可以成功地应用在医疗文本分类领域。

A Medical Text Classification Model Based on Meta-learning

Medical texts have complex technical terms, insufficient training samples in vertical fields, and traditional classification methods cannot meet the needs, this paper proposes a few-shot text classification model to improve medical texts classification. This model is based on transfer learning, adding attention mechanism to give different weights to different words in the sentence, using two competing neural networks to play the roles of domain recognizer and meta-knowledge generator respectively, and strengthening meta-learning adaptation to new dataset through adaptive network. using ridge regression to obtain the classification of the dataset. The comparative experimented results and analysis verify that this model has a good classification on some public text datasets and medical texts. A few-shot text classification model based on meta-learning can be successfully applied in the medical text classification.