Machine Learning & AI


Machine Learning is an ensemble of statistical techniques that give computers the ability to learn using large amount of data, without being explicitly programmed.

Data > Prediction & knowledge > Action

Imagine to write a program that can read digits you wrote on a piece of paper. In traditional programming, you would need a lot of “If - Then - Else”, e.g. “If there’s a circle and a stick below on the right hand side, then it’s a 9.”. Programming this would already be challenging! Worst, it would probably only work for your hand-writting and for sure only for this particular task.


Machine Learning takes a different approach: in this example, you would use an all-purpose well-known algorithm and feed it with labeled data (i.e. pictures of digits for which we know the meaning). The chosen algorithm will learn from the training data and make predictions on test data (to measure accuracy, Sensitivity & Specificity) or on new data. That same algorithm could be used for very different tasks, using very different data.

This technology can make predictions (e.g. “Is this email a spam?”, “Is there a tumor on this MRI?”) or gain knowledge from your data (“Customers that buy X tend to buy Y.”). It can solve very complex business problems and automate work. It creates a huge disruption in many industries.


Effective Machine Learning can be challenging to implement and often fails to deliver the expected value. You need the right data, algorithms, people and questions. If you throw garbage in, you will get garbage out. And even when the planets align, bad mistakes can happen so you need to be careful.

Machine Learning works as a black box. The benefit is that you do not need to understand completly the algorithm used (although it can help), and therefore the learning curve is reasonable. Downside: it is often difficult to understand why a prediction was made.

Engage with this technology! It is rocket science that everybody can use. It can clearly solve problems you have that would be very challenging to tackle with traditional programming.

Educating for the future is a PhUSE working group dedicated to gather and curate educational material on several topics, such as Machine Learning. You can actually join us and help us educate the industry on this exciting topic!

Learn more