Zementis is a decision engine for operational deployment, management and execution of AI and Machine Learning models. Zementis comes in three major flavors and is available as a stand-alone platform in the cloud and on-prem, as a microservice application in Cumulocity IoT and as a plugin component on top of various Hadoop-based data storage platforms and stream processing engines.
The Predictive Analytics microservice in Cumulocity IoT now comes with:
- Codeless device data and ML model interface mapping
- Scheduling of batch prediction jobs
- Zementis engine in all flavors now offers:
- Enhanced support for Time Series Models
- Support for Seasonal ARIMA forecasting
- Enhanced support for CNN architectures like MobileNet, VGGNet, ResNet and RetinaNet.
- Support for Object Localization along with object detection
- Compliance with latest release of PMML 4.4 including:
- Support for Lag expressions with aggregate functions
- Support for Anomaly Detection Models
- Enhanced support for Time Series Models
Open Source releases:
- pmml and pmmlTransformations R packages now available on GitHub
- first release of the cumulocityr R package for extracting data from Cumulocity’s database
- 4.0 release of Nyoka and updated documentation (Redirecting to https://softwareag.github.io/nyoka/)
Release Notes are available at:
http://techcommunity.softwareag.com/ecosystem/communities/public/webmethods/.links/documentation-zementis