Overfitting CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1expert · hard

   A team is training a model to classify medical images. They have a small dataset of 1,000 images per class. The model is a deep convolutional neural network. They observe overfitting. Which of the following is the most effective way to address overfitting while preserving the model's ability to learn relevant features?

   • AIncrease the learning rate to speed up training.
   • BReduce the number of layers in the network to make it simpler.
   • CRemove the validation set and use all data for training.
   • DUse data augmentation to artificially increase the size of the training set.
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A team is training a deep neural network for a classification task. They have a large dataset but notice that the model overfits. They are considering using early stopping, L2 regularization, and dropout. Which of the following statements is true?

   • AL2 regularization increases the bias of the model, which always leads to underfitting.
   • BDropout is only effective for convolutional neural networks.
   • CEarly stopping is the only technique that can prevent overfitting.
   • DL2 regularization and dropout both reduce overfitting by constraining the model's capacity.
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which of the following is a technique to prevent overfitting?

   • ARemoving the validation set from the training process.
   • BIncreasing the number of training epochs without monitoring validation performance.
   • CUsing a more complex model with more parameters.
   • DEarly stopping
   Reveal Solution Discussion Select an answer first
5. 4expert · hard

   A data scientist is training a model and observes that both training and validation errors are high. The model is a simple linear regression. Which of the following is the most appropriate next step?

   • AUse early stopping to halt training sooner.
   • BReduce the number of features to simplify the model.
   • CIncrease the model complexity, for example by adding polynomial features.
   • DApply stronger regularization to reduce overfitting.
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A data scientist is training a logistic regression model on a dataset with 50 features and only 300 samples. The model shows signs of overfitting. Which of the following techniques is most likely to reduce overfitting in this situation?

   • AUse a more complex model such as a deep neural network.
   • BApply L2 regularization to penalize large coefficients.
   • CRemove the regularization term to allow the model to fit the data more closely.
   • DIncrease the number of features to give the model more information.
   Reveal Solution Discussion Select an answer first