Starts
Ends
(Timezone - Europe/Amsterdam)
Experimental Design in Machine Learning for Research
Description
This module emphasises machine learning as a research methodology rather than as an implementation assignment. Participants examine the translation of research objectives into machine learning problem formulations, the selection and evaluation of models, and the interpretation and reporting of results in a scientifically rigorous manner. Scenarios address common research issues, including data leakage, overfitting, and misinterpretation of performance metrics.
A dedicated focus addresses distributed patterns for data confidentiality: how Azure ML can train models on data that cannot leave the customer's on-premises environment. Azure Confidential Computing, Azure
Arc-enabled ML, and federated learning patterns are presented as practical solutions for institutions bound by GDPR, health data directives, or institutional data governance policies.
This module serves as an application-focused extension connecting core cloud literacy with everyday research practices, strengthening programme objectives around scientific rigour, reproducibility, and ethical use of AI while establishing Azure as an integrated research ecosystem.
![]()
Learning Outcomes
Set up an Azure Machine Learning workspace and configure compute for research experiments
- Design and run ML experiments with full tracking using MLflow for parameters, metrics, and artefacts
- Apply hyperparameter tuning and AutoML to systematically optimise models for research tasks
- Identify which distributed/confidential computing pattern applies to specific data residency and privacy requirements
- Deploy trained models to real-time and batch endpoints with monitoring for production research use
- Integrate Azure ML with Fabric and Foundry for end-to-end research AI workflows
Registration
Participants
The agenda of this meeting is empty