Machine Learning Classification 5

Question 1

What is 'label smoothing' in classification training?

Accepted Answer

Replacing hard 0/1 targets with soft values (e.g., 0.1/0.9) to prevent overconfident predictions

Answer

Label smoothing replaces hard class labels with softer distributions, regularizing the model and preventing it from becoming overconfident on training examples.

Question 2

Which phenomenon occurs when a classifier performs well on training data but fails to generalize to unseen examples?

Accepted Answer

Overfitting

Answer

Overfitting means the model has memorized training patterns including noise, resulting in high training accuracy but poor generalization to new data.

Question 3

In Naive Bayes classification, what 'naive' assumption is made?

Accepted Answer

Features are conditionally independent given the class label

Answer

Naive Bayes assumes that all features are conditionally independent given the class, allowing the joint likelihood to be factored into a product of individual likelihoods.

Question 4

What is the primary advantage of using an ensemble method like Gradient Boosting over a single decision tree?

Accepted Answer

Higher predictive accuracy by combining many weak learners sequentially

Answer

Gradient Boosting combines many shallow trees sequentially, each correcting previous errors, typically achieving much higher accuracy than any single tree.

Question 5

Which of the following best describes 'calibration' of a classifier?

Accepted Answer

Ensuring predicted probabilities align with actual observed frequencies

Answer

A well-calibrated classifier produces probability estimates that match empirical outcomes; for example, 70% predicted probability should correspond to 70% actual positive rate.

Machine Learning Practice Test

Machine Learning Practice Test

Machine Learning Classification 5