This video is a summary of Timo Fruytier and Usman Zafar their talk as presented during the previous software meetup (www.meetup.com/prodrive-techn....
Timo Fruytier will discuss his research project on combining multiple manufacturing data streams to predict the success rate of a product during manufacturing. Predicting the success rate can save component costs, and engineers are guided in their scrap root cause analysis.
Usman Zafar will share how to approach the subject of monitoring in a more structured way. Deploying machine learning models to a live production environment where they can deliver business value is a big achievement and not often an easy task. However, the deployment is not the end of the road, it is the beginning. Once deployed and operating in the real world, a model can become vulnerable to the changing data that will be input into the model. Data is not static, but dynamic and evolving. The value of the training data we use to train a model, will decay over time as the world around us changes. As deploying models becomes easier and more widespread, monitoring will become increasingly more important so that they continue to drive value for businesses.
Interested to join our next meetup? Head over to meetup.com to sign up! www.meetup.com/prodrive-techn...