Edge AI

We build models that run on small, low-power devices, so the device decides on site and sends an event rather than a stream of raw data. Sensitive data stays where it was collected, and less has to travel, be stored or be secured.

Channel AChannel BChannel CScoreRaw inDevicemodel runson siteEvents outE1E2E3
Channel AChannel BChannel CScoreRaw inDevicemodel runson siteEvents outE1E2E3
0.70
Model score, score strip
0.0 normal1.0 anomalous

720 raw values in. 0 raw values out. 3 event records out, at a score of 0.70 or more.

  1. E1samples 60 to 62, Channel A0.87
  2. E2samples 131 to 134, Channel B0.82
  3. E3samples 203 to 205, Channel A0.83
View 3 channels, 240 samples each Units model score, 0 to 1 Source synthetic signals