Zementis is a decision engine for operational deployment, integration and execution of AI and machine learning models. With the 10.2 release we have added Deep Learning support for the first time ever to Zementis based on a new representation of Deep Neural Networks in PMML.
- Support for Deep Learning, specifically Convolutional Neural Network (CNN) architectures including LeNet, MobileNet and VGGNet
- New look & feel for Zementis REST UI and updated Zementis REST API to OpenAPI Specification 2.0
- New Python library Nyoka to export PMML from Python-based ML/DL environments like Scikit-learn, Keras, Tensorflow or Caffe
- Replaces earlier versions of Python-based Zementis exporters (Py2Pmml)
- Supports 15 ML algorithms
- Supports Deep Neural Networks here in particular CNNs in a new, very compact version inside PMML in one file
- Supports writing customized exporters using the complete definition of the latest PMML schema plus extensions for Deep Neural Networks
Release Notes are available at the following URL