Easy Domain Adaptation for cross-subject multi-view emotion recognition

作者：

Highlights：

• EasyDA achieves encouraging accuracy in SEED and SEED-IV datasets compared to competitive domain adaptation methods.

• EasyDA dramatically reduces the computational requirement compared to multiple representative domain adaptation methods.

• EasyDA eliminates the exhaustive hyperparameter tunings in domain adaptation, which is simple to implement and use.

摘要

•EasyDA achieves encouraging accuracy in SEED and SEED-IV datasets compared to competitive domain adaptation methods.•EasyDA dramatically reduces the computational requirement compared to multiple representative domain adaptation methods.•EasyDA eliminates the exhaustive hyperparameter tunings in domain adaptation, which is simple to implement and use.

论文关键词：Emotion recognition,Domain adaptation,Approximate empirical kernel map,Distribution alignment,Manifold regularization

论文评审过程：Received 20 August 2021, Revised 1 November 2021, Accepted 13 December 2021, Available online 21 December 2021, Version of Record 5 January 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107982