MICROSOFT AZURE ML - Ops
This article aims at the proceeding of how to use Microsoft Azure Machine Learning Studio to manage the lifecycle of your own models. The very Azure Machine Learning uses ML-Ops also known as Machine Learning Operations which enhances the quality and consistency of every Machine Learning model (ML - models).
Azure ML provides the following ML-Ops capabilities :
- Deploy Machine Learning ML projects from anywhere
- Monitor Machine Learning ML for operational and ML related issues
- Automate the end to end Machine Learning Lifecycle with Azure Machine Learning and Azure DevOps
- Capture the data required for establishing an end to end audit trail to the Machine Learning lifecycle
Turn your training process into a reproducible pipeline through azure services
Using Machine Learning ML pipelines from Azure to stitch all the steps involved in the model training process, from data preparation to feature extraction to hyperparameter turning to model evaluation.
For more detailed info click here
Monitor Machine Learning ML for Operational
Monitoring enables you to understand the data being sent to your model created, and the predictions that it returns. Thus, this information helps to understand the model created. The collected input data may also be useful in a training version of future models.
Audit trail of the Machine Learning Lifecycle
Azure Machine Learning ML gives the freedom to track end to an end audit trail of all your Machine Learning assets, these might be :
- Azure ML Datasets can be helpful to track and the version of the data.
- Azure Machine Learning ML integrates with Git to track information
- Azure Run history track all history the code, data