Try this exercise. Click the correct answer from the options.
Which change most directly combats overfitting?
Click the option that best answers the question.
- Increase learning rate dramatically
- Add L2 penalty and use early stopping
- Remove validation set
- Train forever