Lak, Sara and Michael will introduce three of these tried-and-proven methods to help engineers tackle problems that frequently crop up during the ML process. And we will wrap up the event with AMA (Ask Me Anything).

There will be a Kahoot quiz with top 5 winners getting a free copy of the book.

AGENDA

* 5:00pm Talk by Lak

* 5:15pm Talk by Sara

* 5:30pm Talk by Michael

* 5:45pm AMA (Ask Me Anything)

TALKS

The Bridged Schema pattern, Valliappa Lakshmanan

When an input provider makes improvements to their data feed, it often takes time for enough data of the improved schema to be collected for us to adequately train a replacement model. The Bridged Schema pattern allows us to use as much of the newer data as is available, but augment it with some of the older data to improve model accuracy.

The Rebalancing pattern, Sara Robinson

Many real-world datasets are not perfectly balanced, and it’s important to address this throughout the ML process – in data analysis, model development, and in production. The Rebalancing pattern provides various approaches for handling datasets that are inherently imbalanced.

The Continued Model Evaluation pattern, Michale Munn

This pattern handles the common problem of detecting when a deployed model is no longer fit-for-purpose. This pattern addresses the problems of data and concept drift by regularly evaluating your model, and using these results to determine if retraining is necessary.