Overfitting CT-AI Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A data scientist is training a neural network for image classification. The training loss decreases rapidly, but the validation loss starts to increase after a few epochs. Which of the following actions is most likely to improve validation performance?

   • AIncrease the number of epochs to allow the model to converge further.
   • BAdd dropout layers to the network.
   • CIncrease the learning rate to speed up training.
   • DRemove the validation set and train on all data.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A data scientist trains a model to predict customer churn. On the training dataset, the model achieves 99% accuracy, but on a held-out test set it achieves only 62% accuracy. Which of the following best describes what is happening?

   • AThe model is performing correctly because high accuracy on training data is the primary goal.
   • BThe model is overfitting because it has memorized the training data and fails to generalize to new data.
   • CThe model has a high bias problem, which is causing it to ignore important features.
   • DThe model is underfitting because it is too simple to capture the patterns in the data.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A model trained to detect fraudulent transactions achieves 99.9% accuracy on the training data but only 70% accuracy on a new batch of transactions. Which of the following is the most likely explanation?

   • AThe model has learned the training data too well, including noise, and fails to generalize.
   • BThe model is underfitting because it is not complex enough.
   • CThe model has high bias and is ignoring important features.
   • DThe new batch of transactions is from a different distribution, which is unrelated to overfitting.
   Reveal Solution Discussion Select an answer first
5. 19expert · hard

   A data science team is developing a model to predict equipment failure. They have a small dataset of 500 records and are using a random forest. The model shows high training accuracy but poor validation accuracy. The team is considering several actions. Which action is most likely to improve validation accuracy without sacrificing too much training performance?

   • AIncrease the number of features considered at each split.
   • BRemove the validation set and use the entire dataset for training.
   • CIncrease the number of trees in the forest.
   • DDecrease the maximum depth of each tree.
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   Which technique is commonly used to prevent overfitting by adding a penalty to the model's complexity?

   • ARegularization
   • BCross-validation
   • CData augmentation
   • DEarly stopping
   Reveal Solution Discussion Select an answer first