The International Conference on Machine Learning 2015 in Lille, France was a fantastic opportunity for us to meet the world’s best ML researchers! Many of the conversations we had on the Microsoft booth were around using Azure Machine Learning (#AzureML) for teaching, so we thought we would expand on those discussions here.
Effectively conveying the foundations and practical aspects of machine learning to undergraduate and Masters-level students can be a challenge, as this is such a huge field. Providing simple but exciting hands-on learning experiences for students to explore the what, why and how of ML without getting bogged down in technical minutiae early on is not easy. The browser-based Azure ML Studio can really help, and without students even needing any experience in computer programming.
For courses where students are meeting machine learning for the first-time, exciting them with simple, but powerful, examples of how ML can predict the future right off the bat can be a great start to their learning journey. Azure ML provides a very low-friction way of enabling students to discover how different ML algorithms perform using real-world examples, such as predicting car prices, estimating Twitter sentiment, Detecting credit risk anomalies and predicting flight delays.
Where Azure ML really helps is through instructors pre-building ML workflows and sharing them with students, either privately using a collaborative workspace, or publicly through the Azure ML gallery. As an introduction, you can then let them explore these models and, for example, explain the process of evaluating models using ROC, AUC, and other metrics in a hands-on way.
(Figure shows evaluating a model in Azure ML, including ROC curve and scoring metrics – see https://azure.microsoft.com/en-us/documentation/articles/machine-learning-walkthrough-4-train-and-evaluate-models/)
You can setup tutorial workflows with data and allow students to ‘fill-in-the-blanks’, maybe comparing how different ML algorithms perform on the same problem. There is a plethora of sample datasets built into ML Studio for you to create educational material around, as well as many tutorials already built by the community published in the gallery. Once students are familiar with Azure ML they can start to use it for their own projects, working individually or in teams. A unique feature is the ability to include your own R and Python code, so there is ultimate flexibility in what you can do. And when a model has been validated, it is easy to publish this as a web service with an auto-documented REST API, to be consumed by apps.
We’re excited to see how you can use Azure ML in your own learning and teaching scenarios, not just at universities, but in companies, schools and anywhere! There are three easy ways you can get started with Azure ML for education:
- There is a free tier that includes 10GB of Azure storage for our datasets, and ability to build Azure ML experiments for an hour with up to 100 modules. Get started with this here.
- Azure for Education is for Faculty running courses using Azure, including Azure ML. Each student receives $100 of Azure credit per month, for 6 months. The Faculty member receives $250 per month, for 12 months. You can apply anytime at http://www.microsoftazurepass.com/azureu
- Azure Machine Learning for Research is for University Faculty running data science courses who may need greater amounts of Azure storage and additional services such as HDInsight (Hadoop and Spark) or DocumentDB (NoSQL). Proposals are accepted every two months, you can find out more and apply at http://research.microsoft.com/en-us/projects/azure/ml.aspx
We hope that Azure Machine Learning can help us train machine learning researchers, practitioners and data scientists together. As with many organisations, Microsoft has a growing need for smart talent in this area, so we fully support the efforts of academia to educate and inspire the next-generation.
So please explore and let us know what you are doing with Azure Machine Learning and your students, we’re here to help.
Dr Kenji Takeda, Microsoft Research
kenji.takeda <at> microsoft.com