aberration

Continuous anomaly detectionFor connected systems

Evidence
of change.

Every system has a rhythm. Aberration follows the relationships in your sensor data and brings the moments of change into view.

Open dashboard How it works
A sensor stream leaving its calibrated band A single sensor trace runs inside a shaded reference band. At the marked change point it drifts out of the band and turns teal; the points outside the band are highlighted. τ CHANGE POINT evidence ≥ 1/α t − 60 s t − 40 s t − 20 s t
Fig. 01A stream leaves the band its baseline was calibrated to. Evidence accumulates with every observation until the change is flagged.
Input
MQTT telemetry
Processing
Streaming pipelines
Calibration
Conformal scores
Output
Reviewable evidence
How it works

See the change in context. From connected sensor streams to evidence you can explore.

  1. 01 Observe

    Read the relationships.

    Bring your sensor streams together. See how measurements behave in relation to one another, beyond individual readings.

  2. 02 Calibrate

    Establish the reference.

    Learn from baseline data and calibrate anomaly scores with conformal methods. Give new observations a meaningful reference.

  3. 03 Detect

    Follow the evidence.

    Accumulate evidence as observations arrive. Explore detected changes alongside the telemetry that explains them.

Start

Your systems, in perspective.

Explore your streams and follow the evidence.

Open dashboard