Skip to content

Structured Convolutional Projection

Home / Concepts / Structured Convolutional Projection

Structured Convolutional Projection

Auto-generated stub. Edit this file to add more details.

A technique for projecting multivariate inertial sensor data into the latent space of LLMs for downstream activity recognition.

Why It Matters

It provides a novel methodology for bridging the modality gap between heterogeneous inertial time series data and frozen LLM backbones.

Evidence

we introduce a structured convolutional projection that maps multivariate accelerometer and gyroscope signals into the latent space of the LLM.

Metadata & Links