Thanks to the rise of wearable and connected devices, sensor-generated timeseries comprise a large and growing fraction of the world's data.Unfortunately, extracting value from this data can be challenging, sincesensors report low-level signals (e.g., acceleration), not the high-levelevents that are typically of interest (e.g., gestures). We introduce atechnique to bridge this gap by automatically extracting examples of real-worldevents in low-level data, given only a rough estimate of when these events havetaken place. By identifying sets of features that repeat in the same temporal arrangement,we isolate examples of such diverse events as human actions, power consumptionpatterns, and spoken words with up to 96% precision and recall. Our method isfast enough to run in real time and assumes only minimal knowledge of whichvariables are relevant or the lengths of events. Our evaluation uses numerouspublicly available datasets and over 1 million samples of manually labeledsensor data.
展开▼
