Journal of Computer Applications ›› 2018, Vol. 38 ›› Issue (9): 2483-2488.DOI: 10.11772/j.issn.1001-9081.2018020440

Previous Articles     Next Articles

Deep automatic sleep staging model using synthetic minority technique

JIN Huanhuan1, YIN Haibo2, HE Lingna1   

  1. 1. College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou Zhejiang 310023, China;
    2. School of Astronautics, Harbin Institute of Technology, Harbin Heilongjiang 150001, China
  • Received:2018-03-07 Revised:2018-05-09 Online:2018-09-10 Published:2018-09-06
  • Contact: 何玲娜
  • Supported by:
    This work is partially supported by the Technology Program Public Technology Project of Zhejiang Province (2015C31111).

基于生成少数类技术的深度自动睡眠分期模型

金欢欢1, 尹海波2, 何玲娜1   

  1. 1. 浙江工业大学 计算机科学与技术学院, 杭州 310023;
    2. 哈尔滨工业大学 航天学院, 哈尔滨 150001
  • 通讯作者: 何玲娜
  • 作者简介:金欢欢(1990—),女,安徽亳州人,硕士研究生,主要研究方向:机器学习、脑机接口、时序信号处理;尹海波(1990—),男,安徽亳州人,硕士研究生,主要研究方向:智能诊断、人工神经网络、机器学习;何玲娜(1978—),女,浙江杭州人,副教授,硕士,主要研究方向:脑机接口、人工智能。
  • 基金资助:
    浙江科技计划公益技术项目(2015C31111)。

Abstract: Since current available sleep electroencephalogram data sets for sleep staging are all class imbalanced small data sets, it is hard to achieve ideal staging result by directly migration application of deep learning models. A deep automatic sleep staging model for class imbalanced small data sets was proposed, from the aspect of data oversampling and model training optimization. Firstly, a Modified Synthetic Minority Oversampling TEchnique (MSMOTE) was improved from the perspective of reducing the decision region, and the new technique was applied to generate the minority class samples in the original data sets. Then, the reconstructed class balanced data sets were used to pre-activate the sleep staging model. The 15-fold cross-validation experiment showed the overall classification accuracy was 86.73% and the macro-averaged F1-score was 81.70%. The value of F1 for the minimum class increased from 45.16% to 53.64% by using the data sets reconstructed by improved MSMOTE, to pre-activate the model. In conclusion, the model can realize the end-to-end learning for raw sleep electroencephalogram signals. It has a higher classification efficiency by comparison with recent advanced research and is suitable for the portable sleep monitors that work in conjunction with remote servers.

Key words: deep learning, oversampling, residual connection, sleep staging, transfer learning

摘要: 针对现阶段可用睡眠脑电数据皆为类不平衡小数据集,深度学习模型的直接迁移应用所取得的分期效果较差的问题,分别从数据集重构和模型训练优化两方面入手,提出可用于少量类不均衡原始睡眠脑电数据集的深度自动睡眠分期模型。首先,从减少决策域的角度对修改的生成少数类过采样技术(MSMOTE)进行改进,并将其用于数据集中少数类的生成;然后,用重构后的数据集对模型作预激活处理。15折交叉验证得出总体精度和宏F1值分别为86.73%和81.70%。应用改进后的MSMOTE重构的数据集对模型作预激活,可使最小类的F1值由45.16%增至53.64%。实验表明,模型可实现对少量原始睡眠脑电数据的端到端学习,总体分类效果优于近年高水平模型,适用于配备远程服务器的分体式便携睡眠监测设备。

关键词: 深度学习, 过采样, 残差连接, 睡眠分期, 迁移学习

CLC Number: