How can unit testing help mitigate issues associated with machine learning models that are incorporated into software applications?

$1$ point

Unit tests can validate sets of code every time new code is pushed to the application.

Unit tests can periodically retrain a model on new data.

Unit tests can evaluate the strength of access controls in simulated attacks.

Unit tests can validate the integrity of data at each stage of a pipeline.