Total points 3
1.
Question 1
Which ones of the following features are Responsible AI factors? (Select all that apply)

1 / 1 point

> Fairness

Correct
That’s right! Fairness is an essential factor to Responsible AI. The aim is to work towards fair and inclusive systems for all users.


Transparency


> Privacy

Correct
That's right! Privacy is part of Responsible AI since training models, often using sensitive data, need privacy-preserving safeguards.


> Interpretability 

Correct
Correct! Interpretability is crucial for understanding how and why ML models make specific predictions and results. It can inevitably affect customers and other stakeholders' behavior, as they can question or challenge the model's decisions.

2.
Question 2
DNNs can be fooled into misclassifying inputs with no resemblance to the true category. Regarding image classification, this can be done by:

1 / 1 point

Examples Imputation


Dimension Reduction


Partial Deletion. 


> Generating adversarial images.

Correct
Spot on! The network’s classification of any image into any target class can be changed by making mostly imperceptible alterations to the pixels.

3.
Question 3
If an ML model was trained with many examples from a particular group, _________.

1 / 1 point

There would likely be less certainty about its results for examples from different groups.


> There would likely be greater certainty about its results for that group’s examples.


There would likely be less certainty about its results for that group’s examples.


There would likely be greater certainty about its results for examples from different groups.

Correct
That's right! If a model is trained with a big enough number of examples from a particular group, there will be higher certainty about its results for that group’s examples.