Underfitting CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A team is using a deep neural network to classify images. The model achieves 85% accuracy on the training set and 84% on the validation set, but the target is 95%. The team suspects the model is underfitting. Which change is most likely to help?

   • AIncrease the dropout rate to prevent overfitting.
   • BReduce the number of training epochs to avoid overfitting.
   • CAdd L2 regularization with a higher penalty coefficient.
   • DIncrease the number of layers and neurons in the network.
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A data scientist is training a model to predict equipment failure. The model achieves 99% accuracy on the training set but only 55% accuracy on the test set. The classes are highly imbalanced (95% no-failure, 5% failure). The data scientist initially suspects overfitting. However, after reviewing the learning curves, they see that both training and validation accuracy are high and flat. What is the most likely issue?

   • AThe model is overfitting because the test accuracy is much lower than the training accuracy.
   • BThe model is well-fitted because the training accuracy is high.
   • CThe model is not overfitting; the high accuracy is misleading due to class imbalance, and the model may be underfitting the minority class.
   • DThe model is underfitting because the training accuracy is not 100%.
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which action is most likely to help reduce underfitting?

   • AAdding more relevant features to the model.
   • BReducing the model's complexity.
   • CUsing a smaller training dataset.
   • DIncreasing the regularization strength.
   Reveal Solution Discussion Select an answer first
5. 4expert · hard

   A data scientist is training a model and observes that the training loss is decreasing, but the validation loss is increasing. They suspect overfitting and apply early stopping. After early stopping, both training and validation loss are high. What is the most likely issue?

   • AThe model was overfitting, and early stopping stopped training too early, leaving the model underfitted.
   • BThe model was well-fitted, and early stopping had no effect.
   • CThe model was underfitting, and early stopping correctly prevented overfitting.
   • DThe learning rate was too high, causing the model to diverge.
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A team is using a neural network for image classification. The model is underfitting, with 70% accuracy on both training and validation sets. The team decides to increase the model's capacity by adding more layers. After the change, the training accuracy rises to 98%, but the validation accuracy drops to 65%. What should the team do next?

   • AReduce the learning rate to help the model converge better.
   • BIncrease the number of layers further to improve the training accuracy to 100%.
   • CAdd regularization, such as dropout or L2, to reduce the overfitting caused by the increased capacity.
   • DRevert to the original smaller model because the larger model is overfitting.
   Reveal Solution Discussion Select an answer first