TR2020-128

Disentangled Adversarial Autoencoder for Subject-Invariant Physiological Feature Extraction


    •  Han, M., Ozdenizci, O., Wang, Y., Koike-Akino, T., Erdogmus, D., "Disentangled Adversarial Autoencoder for Subject-Invariant Physiological Feature Extraction", IEEE Signal Processing Letters, DOI: 10.1109/LSP.2020.3020215, Vol. 27, pp. 1565-1569, September 2020.
      BibTeX TR2020-128 PDF
      • @article{Han2020sep,
      • author = {Han, Mo and Ozdenizci, Ozan and Wang, Ye and Koike-Akino, Toshiaki and Erdogmus, Deniz},
      • title = {Disentangled Adversarial Autoencoder for Subject-Invariant Physiological Feature Extraction},
      • journal = {IEEE Signal Processing Letters},
      • year = 2020,
      • volume = 27,
      • pages = {1565--1569},
      • month = sep,
      • doi = {10.1109/LSP.2020.3020215},
      • issn = {1558-2361},
      • url = {https://www.merl.com/publications/TR2020-128}
      • }
  • MERL Contacts:
  • Research Areas:

    Artificial Intelligence, Machine Learning, Optimization, Signal Processing

Recent developments in biosignal processing have enabled users to exploit their physiological status for manipulating devices in a reliable and safe manner. One major challenge of physiological sensing lies in the variability of biosignals across different users and tasks. To address this issue, we propose an adversarial feature extractor for transfer learning to exploit disentangled universal representations. We consider the trade-off between task-relevant features and user-discriminative information by introducing additional adversary and nuisance networks in order to manipulate the latent representations such that the learned feature extractor is applicable to unknown users and various tasks. Results on cross-subject transfer evaluations exhibit the benefits of the proposed framework, with up to 8.8% improvement in average accuracy of classification, and demonstrate adaptability to a broader range of subjects.