Mislabeled Data in Datasets CT-AI Practice Questions (Page 5)

Questions 21–24

1. 21foundation · easy

   What is a potential impact of mislabeled data on model evaluation?

   • AIt only affects the training loss, not the evaluation metrics
   • BIt has no impact on evaluation metrics because they are computed on predictions
   • CIt can lead to an overestimation or underestimation of model performance
   • DIt automatically causes the model to fail all test cases
   Reveal Solution Discussion Select an answer first
2. 22foundation · easy

   What is a robust learning technique that can handle mislabeled data?

   • ATraining a model with a loss function that is robust to label noise
   • BIncreasing the number of training epochs
   • CUsing a larger batch size during training
   • DApplying feature scaling to all features
   Reveal Solution Discussion Select an answer first
4. 23foundation · easy

   In the context of ML datasets, what is the defining characteristic of mislabeled data?

   • AData that is duplicated across the training and test sets
   • BData that contains missing values or null entries
   • CData that contains outliers or extreme values in the features
   • DData whose assigned label does not match the true ground truth
   Reveal Solution Discussion Select an answer first
5. 24foundation · easy

   What is a statistical outlier analysis technique for detecting mislabeled data?

   • ACounting the number of instances in each class
   • BSorting the dataset by label alphabetically
   • CRemoving all instances with missing values
   • DIdentifying instances whose features are far from the class centroid
   Reveal Solution Discussion Select an answer first