Release 10.7.0
Machine Learning Workbench (Build 10.7.0.22.292)
- Multi-modal model building using Jupyter Notebooks, Automated-ML and Neural Network editor.
- Support for transfer learning using popular deep neural network architectures.
- Project-based structure for intuitive encapsulation of data science artifacts - code, data, models, resources.
- Support for project versioning with selective artifacts.
- Model training workflow management for PMML models with in-built task scheduler.
- Model inference pipeline management for ONNX models.
- One-click model deployment to Machine Learning Engine.
Machine Learning Engine (Build 10.7.0.21.11)
- API and UI updated to improve representation of input/output schema for ONNX models.
- Support for latest PMML version 4.4.1.
- Support for ONNX version 1.7.0.
- Nyoka microservice adds a Clustering API to support unsupervised learning on time series data.
- Updated mining field names in PMML generated by Timeseries API to accommodate multi-variate time series in future.