Our Data Science & Technology Meet-Up were absolutely delighted to welcome Pedro Santos for his presentation on Machine Learning for Early Warning Notifications in Industrial Equipment.


Pedro is a Lead Data Scientist at Total Energies UK where he works in the Digital & Data Team who, amongst other key projects, focus on assisting the business to predict when some huge and complex machinery may break down. 


Pedro’s team does this using an unsupervised deep learning approach. Pedro talked us through the ideas behind this approach, what exactly Anomaly Detection was, the steps involved in the process and some of the challenges that the team faced in deployment. 


Watch the full session.

The three main takeaways for us were: 

  1. Predicting failure in the presence of very little labeled data is a difficult problem
  2. There are however elegant solutions to it, which has been proven to work in this field
  3. Digital transformations involve a deep change in behaviors and mindsets for them to be successful

Pedro then fielded some excellent questions from our audience, who were very appreciative of the level of detail and insight that Pedro provided into the inner workings of his team. 


Thank you again, Pedro!