The EM algorithm models the data as being generated by a $q,$

a$.$ mixture of Gaussians.

b$.$ mixture of Sigmoids.

c$.$ means.

d$.$ None of the above.

If a machine learning algorithm is trained on inadequate or inaccurate data, then

a$.$ the algorithm will make bad predictions even if it is itself well designed.

b$.$ the algorithm will make excellent predictions.

c$.$ the algorithm will fix the errors in the data.

d$.$ None of the above.

A runaway trolley is headed toward one person who will be killed. You can hit a swit that will turn the trolley onto an empty sidetrack where there is no one. In this casd clear ethical boundary for machines can be created as:

a$.$ Do nothing $($do not turn the trolley$).$

b$.$ Use the horn to make some noises.

c$.$ Turn the trolley to the empty sidetrack.

d$.$ Shutdown the trolley.

In ML ethics, the key privacy concern is informational privacy and $q,$ be understood as a means to safeguard informational privacy.

a$.$ data gathering.

b$.$ data engineering.

c$.$ data protection.

d$.$ data plotting.

$.$ Current machine learning systems are narrow$/$weak Al because

a$.$ they do not have self$-$ awareness or consciousness.

b$.$ they can think for themselves.

c$.$ they can figure out ethical rules using old dataset without human

d$.$ know if an action is right or wrong.