Latent Sensor Fusion: Multimedia Learning of Physiological Signals for Resource-Constrained Devices
This addresses the problem of efficient biosignal analysis for users of resource-constrained devices, such as in healthcare or wearable technology, and is incremental as it builds on existing latent space and autoencoder techniques.
The paper tackled the computational challenges of analyzing multimodal physiological signals on resource-constrained devices by developing a modality-agnostic, unified encoder using sensor-latent fusion, resulting in a method that is significantly faster, lighter, and more scalable than alternatives without compromising accuracy.
Latent spaces offer an efficient and effective means of summarizing data while implicitly preserving meta-information through relational encoding. We leverage these meta-embeddings to develop a modality-agnostic, unified encoder. Our method employs sensor-latent fusion to analyze and correlate multimodal physiological signals. Using a compressed sensing approach with autoencoder-based latent space fusion, we address the computational challenges of biosignal analysis on resource-constrained devices. Experimental results show that our unified encoder is significantly faster, lighter, and more scalable than modality-specific alternatives, without compromising representational accuracy.