M2 Quiz Answers PDF

Indicate whether the following statements are true or false: 1. Data in ML consists of examples and instances while in MR it is facts and rules. (True) 2. Expert system is an example of ML. (False) 3. Traditionally, a lot of human eCort is needed to develop rules for MR. (True) 4. Normally, goal...

Indicate whether the following statements are true or false: 1. Data in ML consists of examples and instances while in MR it is facts and rules. (True) 2. Expert system is an example of ML. (False) 3. Traditionally, a lot of human eCort is needed to develop rules for MR. (True) 4. Normally, goal of MR is to predict the outcome of the new data. (False) 5. Normally, goal of MR is classification of the data based on its nature. (False) 6. In preparing the data, supervised learning is less complex than unsupervised learning. (True) 7. In preparing evaluating the results, supervised learning is less complex than unsupervised learning. (True) 8. Anomaly detection is a common example of MR. (False) 9. A lot of human eCort is needed to develop rules for ML. (False) 10. Goal of ML is to predict the outcome of thew new data. (True) 11. Goal of ML is classification of the data based on its nature. (True) 12. The goal of RL is to maximise the cumulative reward (True) 13. RL requires human eCort to label the data (False) Question: Mix and match AI techniques with their advantages Description Correct Answer 14. Less human eCort Unsupervised ML 15. Inherently transparent: show the MR mechanism in generating results 16. Able to generate its own policy / RL rules 17.Able to generate accurate Supervised ML prediction Question: Mix and match AI techniques with their disadvantages Description Correct Answer 18. Require long computation for 1. MR large model 19. Safety needs to be ensured 2. RL 20. Data Must be defined in a 3. Supervised ML structured format 21. Require human validation in 4. Unsupervised Learning evaluating result Multiple Choice Questions: 22. An autonomous-driving car can navigate around congested traCic without crashing to another vehicle. Before deployed to the real worlds, this capability is firstly developed in a simulation to minimize the cost of training. This car can identify diCerent type of obstacles such as human or vehicle and follow traCic rules. Q: which AI technique is the most suitable to identify traCic signs or road marks? Correct answer:Classification 23. Consider scenario of question 22. Q: which Trustworthy AI requirement is the most suitable for the overall system? Correct answer: Safety 24. Commercial websites often have advertisement that is curated to the user profile where the preference can be learned by analysing which one is chosen from the suggested advertisement. What AI technique is used for this task? Correct answer: Reinforcement Learning 25. In the scenario above(Q24), what is the most important trustworthy AI aspect during its operation from the user’s perspective? Correct answer: Privacy 26. One way to train a model is to learn from the data collected from diCerent sources. What is the best technique to learn from multiple sources while respecting trustworthy AI aspect? Correct answer: Federated Learning 27. In Generative AI, which of the following is a major challenge? Correct answer: Ensuring that the generated content is factually accurate 28. According to the AI Act, what is not the requirement of implementing high-risk AI system? Correct answer: Open source the implementation to ensure transparency (T) 29. Which one is not considered to be an aspect of a cognitive architecture? Correct answer: Domain-specific facts 30. Which of the following statements correctly describes the relationship between symbolic reasoning and Resource Description Framework (RDF)? Correct answer: RDF is a symbolic representation of data that is used to perform logical inferences using symbolic reasoning techniques 31. Materialization of implicit facts is the output of which machine reasoning method? Correct answer: Forward chaining 32. In a modern telecom system, which technology is essential for enabling true autonomy? Correct answer: Combination of ML and MR 33. What is not considered to be part of the intent that is given to an AI system? Correct answer: Actions