A METHOD FOR EXTRACTING LATENT CONTEXT PATTERNS FROM SENSORS

摘要

A method for extracting latent context from raw data sensor, according to which features are extracted from raw data sensor records by data manipulations, mathematical calculations and extracting statistical features related to sensor data. Latent context patterns are discovered by building a latent context condition applying clustering algorithm and building a latent context attributes applying Deep Learning algorithm and PCA algorithm. Then the discovered patterns are matched into a unified vector which presents the latent context patterns extracted from the raw data sensor.