MLS-C01 Question 188
Select 3You are training a deep learning model on AWS SageMaker for image classification. During training, you notice that your model is overfitting the training data, achieving high accuracy on the training set but poor performance on the validation set. To address this issue, you decide to use dropout. Which of the following statements about dropout are correct?
- A
Dropout randomly sets a fraction of the input units to zero during training to reduce overfitting.
- B
Dropout increases the number of parameters in the model, making it more likely to overfit.
- C
Dropout is only applied during training and is disabled during inference.
- D
Using dropout requires modifying the learning rate to avoid model instability.
- E
Dropout can be combined with other regularization techniques, such as L2 regularization, for better performance.
Show answer and explanation
Correct answers: A, C, E
Explanation
Dropout is a regularization technique used to reduce overfitting in neural networks by randomly deactivating a fraction of neurons during training. This prevents the model from relying too heavily on specific neurons and forces it to learn more robust features. Dropout is only applied during training and is not used during inference. Additionally, it can be used in combination with other regularization techniques like L2 regularization for better generalization.
- A. Correct.
Correct. Dropout works by randomly setting a fraction of the input units to zero during training, which helps prevent the model from relying too heavily on specific neurons and reduces overfitting.
- B. Incorrect.
Incorrect. Dropout does not increase the number of parameters in the model; rather, it temporarily deactivates a fraction of them during training to improve generalization.
- C. Correct.
Correct. Dropout is only applied during training. During inference, dropout is disabled, and the full network is used with scaled weights.
- D. Incorrect.
Incorrect. Dropout does not inherently require modifying the learning rate. While adjusting the learning rate may help in some cases, it is not a requirement for using dropout.
- E. Correct.
Correct. Dropout can be combined with other regularization techniques, such as L2 regularization, to further enhance the model's generalization capabilities.