An ML engineer has trained an ML model by using Amazon SageMaker AI. The ML engineer determines that the model is overfitting and that the training data contains unnecessary features. The ML…

Question

Options

• AIncrease the number of training iterations. Retrain the model.
• BApply L1 regularization to the training data. Retrain the model.
• CDecrease the number of training iterations. Retrain the model.
• DUse SageMaker Debugger to apply L1 regularization to the running model.

• A
  13% (4)
• B
  77% (23)
• C
  3% (1)
• D
  7% (2)

Explanation

The problem has two components: overfitting and unnecessary (irrelevant) features. L1 regularization (Lasso) adds a penalty proportional to the absolute value of model coefficients to the loss function. This has the dual effect of (1) reducing model complexity to combat overfitting, and (2) driving the coefficients of irrelevant features to exactly zero, effectively performing automatic feature selection. L2 (Ridge) regularization shrinks coefficients but rarely eliminates them entirely, so it wouldn't remove unnecessary features. Increasing training iterations (A) would worsen overfitting. Decreasing iterations (C) is a crude fix that doesn't address feature relevance. SageMaker Debugger (D) is a monitoring and debugging tool, not a mechanism to apply regularization to a running model.

No community discussion yet for this question.