Machine Learning Quiz

What can be learned from a predictive model should not change if the excluded in the training set is either included or (SP24_RE) (SU24_RE)

• dataset filter

• biased dataset

• model fairness score

• individual's data

Which of the following is a recommended security practice for machine learning datasets? (SP24_RE) (SU24_RE)

• Ensure your team is full of different perspectives

• Enact a sound data governance structure

• Perform threat modeling with beneficial algorithms

• Create a chain of command

A major downside to k-anonymity is that re-identification is possible with (SU24_RE)

• multiple datasets

• database leaks

• sensitive columns

• expanding k values

A benefit of glass-box models is that if an attribute is skewing the fairness of a decision, a human agent may choose to (SU24_RE)

• trust it

• re-run the algorithm without it

• alert the team

• ignore it

During cross-validation, which of the following dataset splits is used to tune the performance of a model? (SU24_RE)

• Training set

• Test set

• Validation test

• Unit test

When it comes to protecting private data, which of the following could be the root cause of a problem? (SU24_RE)

• The data was stored insecurely.

• The data was leaked online.

• The data was deleted from storage.

• The data was broken.

The Netflix prize privacy scandal is an example of reidentification through (SU24_RE)

• k-anonymity

• database leaks

• multiple datasets

• sensitive columns

If you were to place the task "Apply security patch to system" on an attack tree diagram, what category would it be assigned to? (SU24_RE)

• Threat

• Vector

• Control

• Controller

Which of the following measures is best suited for reporting results to an audience? (SU24_RE)

• Standard deviation

• Variance

• Mode

• Median

Why is it important to take a proactive approach to addressing liability issues? (SU24_RE)

• The law often lags behind innovation.

• The law is not always applied consistently.

• The law is not always clear.

• The law often difficult to understand.