Regardless of the machine learning classifier type chosen, there are many trade-offs involved in how to ensure the model is robust. In some cases we can add regularization or adjust hyperparameters. We can adjust the number of training examples, or use boosting or bagging. This module focuses on techniques for ensuring models are robust.
- Training Robust Models slides.
- Training Robust Models Jupyter notebook.
- Initial release, Susan Davidson and Zachary Ives, University of Pennsylvania, February 2020.