Cumulocity IoT Machine Learning provides enterprises a seamlessly integrated experience for developing IoT machine learning use cases such as predictive maintenance, anomaly detection, demand forecasting etc. The product is based on machine learning standards such as Predictive Model Markup Language (PMML) and Open Neural Network Exchange (ONNX).
[list]
“Machine Learning” replaces the previously named “Predictive Analytics” in Cumulocity IoT.
[/list]
[list]
Available as an add-on for Cumulocity IoT Edge standalone or connected versions.
[/list]
[list]
Open Neural Network Exchange (ONNX), representing Deep Learning models, is supported.
[/list]
[list]
Time series forecasting enhancements include support for multi-variate State Space Models.
[/list]
[list]
Multiple models can be grouped supporting a Champion/Challenger strategy to identify the best performing model in production.
[/list]
[list]
Job scheduling includes pagination to improve performance.
[/list]
[list]
Open Source releases of R-PMML 2.3.0 and Nyoka 4.2.0 support State Space Models for Time Series data.
[/list]
Release Notes are available at: https://cumulocity.com/guides/welcome/intro-documentation/