Research3h ago
New Framework Links Wi-Fi Signals To Logic
A new pipeline uses a categorical variational autoencoder to compress raw Wi-Fi Channel State Information into discrete latent representations. This method extracts Linear Temporal Logic rules to make human activity recognition causally interpretable. It bridges the gap between opaque deep learning and symbolic logic. Practitioners can now audit and modify the specific rules governing activity detection.