This is a dedicated watch page for a single video.
What technique would you use in SageMaker to train a new model using an expanded dataset that contains an underlying pattern that was not accounted for in the previous training and which resulted in poor model performance?