A.  

Maximize precision and accuracy

B.  

Maximize accuracy and recall

C.  

Maximize recall and precision

D.  

Maximize specificity number of classes

A.  

Insufficient data

B.  

Invalid data

C.  

Wanted outliers

D.  

Numbers stored as strings

A.  

The feedback requires a physical connection and cannot be sent over the Internet

B.  

Mobile operating systems cannot process machine learning algorithms

C.  

The size of the application is consuming too much of the phone's storage capacity

D.  

The training, processing, and diagnostic generation are too computationally intensive for the mobile device hardware to handle

A.  

Using the same implementation for data acquisition by data scientists and testers prevents defect masking

B.  

Creating and managing large amounts of test data can be difficult, especially when it needs to be representative

C.  

The input data must always be the same over time, especially in real-world systems

D.  

Artificially generated data requires legal approval and must be sanitized and encrypted

A.  

Threshold coverage

B.  

Neuron coverage

C.  

Sign-change coverage

D.  

Value-change coverage

A.  

Metamorphic testing because the application domain is not clearly understood at this point

B.  

Adversarial testing to verify that no incorrect images have been used in the training

C.  

Pairwise testing using combinatorics to look at a long list of photo parameters

D.  

Back-to-back testing using the version of the model before training and the new version of the model after being trained with additional images

A.  

Selecting training data purposely excluding specific faulty conditions

B.  

Selecting training data by purposely including all known faulty conditions

C.  

Selecting testing data from a different dataset than the training dataset

D.  

Selecting testing data from a boat manufacturer's bolt longevity data

A.  

Internal data labeling

B.  

Crowdsourced data labeling

C.  

Outsourced data labeling

D.  

AI-assisted data labeling

A.  

During the review, it can uncover algorithmic bias by analysing the procedures used to obtain the training data.

B.  

During the review of the preprocessing, the auditor can uncover whether the data has been influenced in a way that could lead to sample distortions.

C.  

It may use the LIME method as part of its data collection review to detect inappropriate bias.

D.  

As part of the review of preprocessing, it can reveal whether the data has been influenced in a way that could lead to algorithmic bias.

A.  

Test automation can be built quickly and easily from the test cases developed during black-box testing

B.  

The tester wishes to better understand the logic of the software used to create the internal structure

C.  

The black-box testing method will allow the tester to check the transparency of the algorithm used to create the internal structure

D.  

Black-box testing eliminates the need for the tester to understand the internal structure of the AI-system

A.  

AI-generated functional test cases typically result in poor requirements coverage.

B.  

A different test oracle is usually required for each AI-generated functional test case.

C.  

Expected results may not be available for AI-generated functional test cases.

D.  

An AI-based system under test must not be used as a functional test oracle.

A.  

Value coverage

B.  

Threshold coverage

C.  

Sign change coverage

D.  

Neuron coverage

A.  

Determinism

B.  

Complexity

C.  

Interpretability

D.  

Robustness

A.  

The bias of a neuron is determined by the activation values of the neurons in the previous layer

B.  

Training a neural network only changes the values of the weights at the connections between neurons

C.  

A single-layer perceptron is NOT a neural network

D.  

The input layer of a deep neural network must have at least as many neurons as its output layer

A.  

Geographical/Locality

B.  

Automation/Complacency

C.  

Complacency/Disregard

D.  

Ignorance/Cognitive

A.  

This statement is incorrect. Current Al is true Al and there is no reason to believe that this fact will change over time.

B.  

This statement is correct. In general, what is considered Al today may change over time.

C.  

This statement is incorrect. What is considered Al today will continue to be Al even as technology evolves and changes.

D.  

This statement is correct. In general, today the term Al is utilized incorrectly.

A.  

Determining the optimal age for a chicken's egg-laying production using input data of the chicken's age and average daily egg production for one million chickens

B.  

Recognizing a knife in carry-on luggage at a security checkpoint in an airport scanner

C.  

Determining if an animal is a pig or a cow based on image recognition

D.  

Predicting shopper purchasing behavior based on the category of shopper and the positioning of promotional displays within a store

A.  

The model suffers from drift and therefore should be regularly tested to ensure that any occurrences of drift are detected soon enough for the problem to be mitigated

B.  

The model suffers from drift and therefore the performance standard should be eased until a new model with more transparency can be developed

C.  

The model suffers from corruption and therefore should be reloaded into the computer system being used, preferably with a method of version control to prevent further changes

D.  

The model suffers from a lack of transparency and therefore should be regularly tested to ensure that any progressive errors are detected soon enough for the problem to be mitigated

A.  

It has been trained to recognize cats and is given an image of a dog

B.  

It has been trained to recognize human faces at a particular resolution and it is given a human face image captured with a higher resolution

C.  

It has been trained to analyze mathematical models and is given a set of landscape pictures to classify

D.  

It has been trained to analyze customer buying trend data and is given information on supplier cost data

A.  

The agent creates a model of the environment from labeled data during training

B.  

The approach is suitable when the application doesnotrequire interaction with the environment

C.  

The agent’s training is based on a reward function that rewards successful attempts

D.  

From experience, the agent learns theoptimal reward function

A.  

AI-based defect prediction is most effective when based on previous similar constellations.

B.  

AI-based defect prediction can detect whether defects exist but not where.

C.  

AI-based defect prediction is most effective when based on source-code metrics such as branches or McCabe complexity.

D.  

AI-based defect prediction is based on formal principles and requires only a few factors.

A.  

Compliance needs require proper care to be taken of input personal data.

B.  

Nature of data constantly changes with lime.

C.  

Data for the use case is being generated at a fast pace.

D.  

Test data being sourced from public sources.

A.  

EDA can help detect the outlier emails from the real emails

B.  

EDA can detect and remove the false emails

C.  

EDA can restrict how many inputs can be provided by unique users

D.  

EDA cannot be used to detect the attack

A.  

Outsourced

B.  

Pre-labeled dataset

C.  

Internal

D.  

Al-Assisted

A.  

Using a pixel comparison of the GUI before and after the change to check the differences.

B.  

Using a computer vision to compare the GUI before and after the test object changes.

C.  

Using a vision-based detection of the GUI layout changes before and after test object changes.

D.  

Using a ML-based classifier to flag if changes in GUI are to be flagged for humans.

A.  

Classification for the optimization of regression test cases

B.  

Probabilistic methods for the prediction of system failures

C.  

Fuzzy logic for the generation of test cases

D.  

Computational optimization techniques for defect prediction

A.  

ML models trained with critical defect tickets can identify defects that cause serious consequences.

B.  

ML models can support duplicate defect identification when checking defect criticality.

C.  

ML models can identify categories for a reported defect during assignment.

D.  

ML models identify developers who should handle a defect based on ticket content.

A.  

The results of analyzing an X-ray for lung cancer using an AI system are more understandable than with a conventional system.

B.  

The X-ray images that an AI system can analyze must be structurally different from X-ray images used in a conventional system.

C.  

An AI system independently determines patterns in X-rays during training; a conventional system requires a human to program in those patterns.

D.  

The implementation of an AI system consists mainly of training data, whereas that of a conventional system consists of branches and loops.

A.  

Recognizing a dog from many different images is a regression problem

B.  

Deciding whether an object is a bicycle or a motorcycle is a classification problem

C.  

Predicting that diesel prices will increase by ~10% is a classification problem

D.  

In classification, objects are always assigned to exactly two classes

A.  

Add 10% of the rows randomly and create another model and compare the R-Square scores of both the model.

B.  

Train various models by changing the order of input features and verify that the R-Square score of these models vary significantly.

C.  

Compare the R-Square score of the model obtained using two different implementations that utilize two different programming languages while using the same algorithm and the same training and testing data.

D.  

Drop 10% of the rows randomly and create another model and compare the R-Square scores of both the models.

A.  

There is no risk, as the model has already been trained

B.  

Insufficient function: the model was not trained to check for colors or words

C.  

Improper data preparation

D.  

Inherited bias: the model could have inherited unknown defects

A.  

The number of parameters to test can be reduced to less than a dozen

B.  

All high priority defects will be identified using this method

C.  

While the number of tests needed can be reduced, there may still be a large enough set of tests that automation will be required to execute all of them

D.  

Pairwise cannot be applied to this problem because there is AI involved and the evolving values may result in unexpected results that cannot be verified

A.  

The model is not able to generalize to accommodate new types of data

B.  

The model is too simplistic for the data

C.  

The model is missing relationships between the inputs and outputs

D.  

The model discards data it considers to be noise or outliers

A.  

Comparison of the results of a current neural network model ML model implemented in platform A (for example Pytorch) with a similar neural network model ML model implemented in platform B (for example Tensorflow), for the same data.

B.  

Comparison of the results of a home-grown neural network model ML model with results in a neural network model implemented in a standard implementation (for example Pytorch) for same data

C.  

Comparison of the results of a neural network ML model with a current decision tree ML model for the same data.

D.  

Comparison of the results of the current neural network ML model on the current data set with a slightly modified data set.

A.  

If only limited data is available, validation and test data sets can be combined in multiple ways during training.

B.  

If limited data is available, it may be better to work without a separate test data set.

C.  

Optimally, the data should be distributed equally between the training, validation, and test data sets.

D.  

The data in the test data set must be equivalent to the data in the training data sets and to the data in the validation data sets.

A.  

Adaptability and flexibility are important when the system needs to change its behavior and determine the change on its own.

B.  

Adaptability is considered to be the ability of the system to be used in unspecified situations.

C.  

Self-learning AI-based systems are classified according to whether they are adaptable only or flexible only.

D.  

Flexibility is considered to be the ease with which the system can be reprogrammed to a changed operating condition.