Federated Remote Physiological Measurement with Imperfect Data

TL;DR

This paper introduces a federated learning system for remote physiological measurement using camera-based sensing, addressing privacy concerns and robustness to noisy data, and demonstrating competitive performance with traditional methods.

Contribution

It presents the first mobile federated learning system for camera-based sensing and proposes a noise-aware weight averaging method to enhance robustness against corrupted data.

Findings

Federated learning achieves competitive results with state-of-the-art methods.

Noise-aware averaging improves model robustness in noisy data scenarios.

The system maintains performance even with low signal-to-noise ratios.

Abstract

The growing need for technology that supports remote healthcare is being acutely highlighted by an aging population and the COVID-19 pandemic. In health-related machine learning applications the ability to learn predictive models without data leaving a private device is attractive, especially when these data might contain features (e.g., photographs or videos of the body) that make identifying a subject trivial and/or the training data volume is large (e.g., uncompressed video). Camera-based remote physiological sensing facilitates scalable and low-cost measurement, but is a prime example of a task that involves analysing high bit-rate videos containing identifiable images and sensitive health information. Federated learning enables privacy-preserving decentralized training which has several properties beneficial for camera-based sensing. We develop the first mobile federated learning camera-based sensing system and show that it can perform competitively with traditional state-of-the-art supervised approaches. However, in the presence of corrupted data (e.g., video or label noise) from a few devices the performance of weight averaging quickly degrades. To address this, we leverage knowledge about the expected noise profile within the video to intelligently adjust how the model weights are averaged on the server. Our results show that this significantly improves upon the robustness of models even when the signal-to-noise ratio is low