Universal Physiological Representation Learning With Soft-Disentangled Rateless Autoencoders

Human computer interaction (HCI) involves a multidisciplinary fusion of technologies, through which the control of external devices could be achieved by monitoring physiological status of users. However, physiological biosignals often vary across users and recording sessions due to unstable physical/mental conditions and task-irrelevant activities. To deal with this challenge, we propose a method of adversarial feature encoding with the concept of a Rateless Autoencoder (RAE), in order to exploit disentangled, nuisance-robust, and universal representations. We achieve a good trade-off between user-specific and task-relevant features by making use of the stochastic disentanglement of the latent representations by adopting additional adversarial networks. The proposed model is applicable to a wider range of unknown users and tasks as well as different classifiers. Results on cross-subject transfer evaluations show the advantages of the proposed framework, with up to an 11. 6% improvement in the average subject-transfer classification accuracy.

Authors

• Mo Han
• Ozan Ozdenizci
• Toshiaki Koike-Akino Mitsubishi Electric Research Labs
• Ye Wang Mitsubishi Electric Research Labs
• Deniz Erdogmus Northeastern University

Keywords

• Feature extraction
• Task analysis
• Physiology
• Decoding
• Biomedical monitoring
• Bioinformatics
• Stochastic processes
• Representation Learning
• Classification Accuracy
• Average Accuracy
• Human-computer Interaction
• Latent Representation
• Improve Classification Accuracy
• Feature Encoder
• Average Classification Accuracy
• Wide Range Of Tasks
• Wide Range Of Users
• Task-relevant Features
• Neural Network
• Training Data
• Classification Task
• Dropout Rate
• Linear Discriminant Analysis
• Feature Dimension
• Transfer Learning
• Generative Adversarial Networks
• Task-related Information
• Latent Features
• Transfer Learning Framework
• Classism
• Nuisance Variables
• Subject ID
• Transfer Learning Method
• Training Data Size
• Adversary Model
• Adversarial Training
• Adversarial learning
• autoencoders
• deep learning
• disentangled representation
• physiological biosignals
• soft disentanglement
• stochastic bottleneck
• Humans
• Machine Learning