Universal Physiological Representation Learning with Soft-Disentangled Rateless Autoencoders

TL;DR

This paper introduces a rateless autoencoder with adversarial feature encoding to learn universal, disentangled physiological representations that are robust across users and tasks, improving cross-subject classification accuracy.

Contribution

The paper proposes a novel adversarial rateless autoencoder framework that effectively disentangles nuisance factors from physiological signals for universal HCI applications.

Findings

Achieves up to 11.6% improvement in cross-subject classification accuracy.

Demonstrates robustness across different users, tasks, and classifiers.

Enables wider applicability of physiological signal analysis in HCI.

Abstract

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.