Lecture 6

Question 1

• What is the need for reducing large adaptation
  spaces in a self-adaptive system?
  1. Make monitoring more efficient.
  2. Make analysis faster.
  3. Make the self-adaptive system more reliable.
  4. Improve the availability of the system.

Question 2

• Which method cannot be used in the situation
  driven optimization mode of the Planning as
  Optimization approach?
  1. Bayesian optimization.
  2. A genetic algorithm.
  3. Local search starting from an initial solution.
  4. Clustering

Question 3

What is the primary reason for reducing large adaptation spaces in self-adaptive systems?
A) To make monitoring more efficient.
B) To make analysis faster.
C) To make the self-adaptive system more reliable.
D) To improve the availability of the system.

Answer 3

B) To make analysis faster.

Question 4

Which of the following is NOT a type of machine learning?
A) Supervised Learning
B) Unsupervised Learning
C) Reinforcement Learning
D) Predictive Learning

Answer 4

D) Predictive Learning

Question 5

What is an adaptation space?
A) The set of all possible adaptation options.
B) The best configuration of the system.
C) The parameters for machine learning models.
D) The output of the MAPE-K loop.

Answer 5

A) The set of all possible adaptation options.

Question 6

Which method is used to determine situations by grouping together environment states?
A) Situation-driven Optimization
B) Clustering
C) Bayesian Optimization
D) Genetic Algorithms

Answer 6

B) Clustering

Question 7

What is a characteristic of reinforcement learning?
A) It requires labeled data for training.
B) It uses a reward function for training.
C) It is used exclusively for classification tasks.
D) It does not require any data for training.

Answer 7

B) It uses a reward function for training.

Question 8

In the optimization of CrowdNav, what is a key challenge?
A) Low dimensionality of configurations.
B) The need for many samples to evaluate a configuration.
C) Lack of known models for inputs and outputs.
D) All configurations are optimal.

Answer 8

B) The need for many samples to evaluate a configuration.

Question 9

Which of the following optimization algorithms is known for multi-objective evolutionary search?
A) Random Search
B) Bayesian Optimization
C) Non-Dominated Sorting Genetic Algorithm II (NSGA-II)
D) Novelty Search

Answer 9

C) Non-Dominated Sorting Genetic Algorithm II (NSGA-II)

Question 10

Which optimization method measures fitness based on a novelty metric?
A) Bayesian Optimization
B) Genetic Algorithms
C) Novelty Search
D) Local Search

Answer 10

C) Novelty Search

Question 11

What is the main assumption of the situation-driven optimization mode?
A) It can be interrupted once started.
B) The optimization process can update the system configuration on the fly.
C) It requires offline training.
D) It uses labeled data for training.

Answer 11

B) The optimization process can update the system configuration on the fly.

Question 12

What is the primary focus of Bayesian optimization?
A) To optimize configurations based on novelty.
B) To perform black-box optimization of continuous spaces.
C) To apply genetic algorithms for optimization.
D) To minimize overhead in high traffic.

Answer 12

B) To perform black-box optimization of continuous spaces.

Question 13

What is a challenge associated with clustering in situation-driven optimization?
A) It is easy to derive and extend.
B) It requires predefined classes.
C) Automated comparison of optimizers.
D) It does not require continuous data collection.

Answer 13

C) Automated comparison of optimizers.

Question 14

Which of the following is a benefit of using machine learning in self-adaptive systems?
A) It makes systems less complex.
B) It enables dealing with large amounts of data.
C) It eliminates the need for runtime decision-making.
D) It reduces the need for monitoring.

Answer 14

B) It enables dealing with large amounts of data.