AIML – PART-A

📘 DOC 1: V20PCA302 - Artificial Intelligence and Machine Learning - Copy.docx

🔹 Q1–Q25 (DOCX — Unit 1 & Unit 2)

Q1. Which of the following is an example of a problem-solving technique in AI?

A) Sorting algorithms
B) Database management
C) Depth-First Search
D) Breadth-First Search

Q2. What is a characteristic of production systems in AI?

A) They are only used in manufacturing industries
B) They consist of rules and facts
C) They are primarily used for entertainment purposes
D) They do not support decision-making processes

Q3. Which of the following is a heuristic search technique?

A) Depth-First Search
B) Binary Search
C) Linear Search
D) Bubble Sort

Q4. What is an intelligent agent?

A) A software program that performs tasks autonomously
B) A person with high IQ
C) A robot with human-like emotions
D) A computer virus

Q5. What is a common implementation of intelligent agents?

A) Chatbots
B) Spreadsheet software
C) Video games
D) Email clients

Q6. In which scenario is Depth-First Search (DFS) particularly useful?

A) When the solution path is known in advance
B) When the search space is infinite
C) When memory usage is a concern
D) When all edge costs are equal

Q7. Which heuristic search technique uses a priority queue to explore nodes with the lowest
estimated cost first?
A) Greedy Best-First Search
B) DFS
C) BFS
D) Hill Climbing

Q8. In which decade did artificial neural networks experience a resurgence in popularity?
A) 1970s
B) 1980s
C) 1990s
D) 2000s

Q9. In which year was the term "Artificial Intelligence" first coined?
A) 1936
B) 1956
C) 1962
D) 1980

Q10. What is the objective of the Travelling Salesman Problem?

A) To find the shortest path between two points
B) To maximize profit while traveling
C) To visit all cities exactly once and return to the starting city
D) To deliver goods to multiple locations

Q11. What is Knowledge Representation in AI?

A) The process of storing information in a computer system
B) The study of how knowledge is stored and manipulated in a computer
C) The process of acquiring knowledge from external sources
D) The study of human cognition and memory

Q12. Which approach to Knowledge Representation represents knowledge using symbolic

structures?
A) Connectionist approach
B) Semantic network approach
C) Frame-based approach
D) Rule-based approach

Q13. What is Predicate Logic used for in Knowledge Representation?

A) To represent relationships between entities using predicates and quantifiers
B) To visualize knowledge using graphical structures
C) To store knowledge in a hierarchical manner
D) To implement neural networks for learning

Q14. Which logic extends propositional logic by including variables, quantifiers, and predicates?
A) First-Order Predicate Logic (FOPL)
B) Modal Logic
C) Temporal Logic
D) Intuitionistic Logic

Q15. What is Knowledge Acquisition in AI?

A) The process of representing knowledge using logical symbols
B) The study of acquiring knowledge from external sources and incorporating it into a
computer system
C) The process of implementing neural networks for learning
D) The study of human cognition and memory

Q16. What is an Ontology in Knowledge Representation?

A) A formal specification of a shared conceptualization
B) A graphical representation of knowledge
C) A set of rules for automated reasoning
D) A database management system

Q17. Which statement best describes the syntax of First-Order Predicate Logic (FOPL)?
A) FOPL uses symbols to represent entities and their relationships
B) FOPL uses graphical structures to represent knowledge
C) FOPL uses rules to represent knowledge
D) FOPL uses natural language to represent knowledge

Q18. What is the primary purpose of representing knowledge using Frames?

A) To organize knowledge into hierarchical structures
B) To enable automated reasoning and decision-making
C) To visualize relationships between entities
D) To implement neural networks

Q19. What is the main advantage of using Predicate Logic for Knowledge Representation?
A) It allows for automated reasoning and inference
B) It provides a graphical representation of knowledge
C) It enables the implementation of neural networks
D) It facilitates natural language processing

Q20. In Predicate Logic, what does a quantifier specify?

A) The number of arguments in a predicate
B) The truth value of a proposition
C) The scope of variables in a logical expression
D) The type of logical connective used

Q21. Which logic-based representation uses “is-a” relationships?

A) Fuzzy logic
B) Predicate logic
C) Semantic networks
D) Modal logic
Q22. Which of the following is not a symbolic representation?
A) Semantic nets
B) Neural networks
C) Frames
D) Rules

Q23. Which of the following is used to represent structured knowledge?

A) Frames
B) Neural nets
C) Rules
D) Probabilities

Q24. In semantic networks, what do arrows represent?

A) Processes
B) Decisions
C) Attributes or relationships
D) Entropy

Q25. Which logic supports variable bindings with quantifiers?

A) Propositional logic
B) Modal logic
C) First-order logic
D) Non-monotonic logic

🔹 Q26–Q50 (DOCX — Unit 3 to Unit 5)

Q26. What is the primary goal of machine learning?

A) To make computers think like humans
B) To enable computers to learn from data and improve performance over time
C) To automate tasks without human intervention
D) To develop intelligent agents for gaming

Q27. Which of the following is NOT a type of machine learning?

A) Supervised Learning
B) Unsupervised Learning
C) Semi-supervised Learning
D) Deterministic Learning

Q28. In supervised learning, what is the role of the input data?

A) To provide the expected output
B) To provide feedback on the model's performance
C) To train the model with labeled examples
D) To generate new features for the model

Q29. What is the task of concept learning?

A) To define new concepts
B) To generalize from given examples
C) To classify data into predefined categories
D) To predict future outcomes

Q30. How is concept learning represented as a search problem?

A) By defining a search space of all possible concepts
B) By minimizing the loss function
C) By optimizing the feature space
D) By maximizing the accuracy of predictions

Q31. Which learning approach learns by interacting with an environment and receiving
feedback?
A) Supervised Learning
B) Unsupervised Learning
C) Reinforcement Learning
D) Semi-supervised Learning

Q32. Which algorithm is used for incremental concept learning in a hypothesis space?
A) ID3 Algorithm
B) Naive Bayes Algorithm
C) Candidate Elimination Algorithm
D) K-Means Algorithm

Q33. What does the version space represent in the candidate elimination algorithm?
A) The set of consistent hypotheses
B) The set of inconsistent hypotheses
C) The set of all possible hypotheses
D) The set of optimal hypotheses

Q34. What is the purpose of implementing the candidate elimination algorithm?

A) To find the most accurate hypothesis
B) To classify data into multiple classes
C) To incrementally learn new concepts
D) To determine the boundaries of the version space

Q35. Which type of learning algorithm is suitable for tasks where the input data is not labeled?
A) Supervised Learning
B) Unsupervised Learning
C) Reinforcement Learning
D) Semi-supervised Learning

Q36. In machine learning, what is the purpose of a decision tree?

A) To classify data points into multiple classes
B) To perform regression analysis on continuous data
C) To visualize the relationship between input and output variables
D) To represent rules for decision-making based on features of the data

Q37. What is the basic algorithm used to build a decision tree?

A) K-Means algorithm
B) ID3 algorithm
C) Logistic regression
D) Naive Bayes algorithm

Q38. What does the hypothesis space search refer to in the context of decision trees?
A) Searching for the best split at each node
B) Searching for the optimal tree structure
C) Searching for the most significant features
D) Searching for the correct class labels

Q39. How does the concept of inductive bias relate to decision trees?
A) It represents the inherent bias in the decision tree algorithm towards certain hypotheses.
B) It represents the bias towards simpler tree structures over complex ones.
C) It refers to the bias towards specific features in the dataset.
D) It refers to the bias towards overfitting the training data.

Q40. What is a common technique used in unsupervised learning for grouping similar data
points together?
A) Clustering
B) Regression
C) Classification
D) Dimensionality reduction

Q41. Which algorithm is commonly used for clustering in unsupervised learning?

A) K-Nearest Neighbors
B) Decision Tree
C) K-Means
D) ID3

Q42. How is entropy calculated in the context of decision trees?

A) As the measure of impurity in a dataset
B) As the average distance between data points
C) As the measure of correlation between variables
D) As the ratio of within-cluster sum of squares to between-cluster sum of squares

Q43. What does the measure of information gain represent in decision trees?
A) The reduction in entropy achieved by splitting the data on a particular attribute
B) The increase in entropy achieved by splitting the data on a particular attribute
C) The accuracy of the decision tree model
D) The complexity of the decision tree model
Q44. What is the primary objective of the K-Means algorithm?
A) To minimize the sum of squared distances between data points and cluster centroids
B) To maximize the within-cluster sum of squares
C) To maximize the between-cluster variance
D) To minimize the number of clusters

Q45. In the context of decision trees, what does the term "version space" refer to?
A) The set of consistent hypotheses
B) The set of inconsistent hypotheses
C) The set of all possible hypotheses
D) The set of optimal hypotheses

Q46. Which algorithm selects the attribute with the highest information gain?
A) ID3
B) Naive Bayes
C) Perceptron
D) Support Vector Machine

Q47. K-Means algorithm falls under which category of learning?

A) Supervised
B) Unsupervised
C) Reinforcement
D) None

Q48. Which of the following is an example of supervised learning?

A) Email spam classification
B) Customer segmentation
C) Topic modeling
D) Anomaly detection

Q49. Which algorithm is used for regression tasks?

A) KNN
B) Linear regression
C) Naive Bayes
D) ID3

Q50. Which method is suitable for non-linear classification problems?

A) Logistic regression
B) Support Vector Machines with kernel trick
C) K-Means clustering
D) Naive Bayes
🔹 Q51–Q75 (JULY 2023 – 25 MCQs)

Q51. Type of matrix decomposition model is:

A) Predictive model
B) Descriptive model
C) Logical model
D) None

Q52. PCA is:

A) Backward feature selection
B) Forward feature selection
C) Feature extraction
D) None of these

Q53. Supervised learning and unsupervised clustering both require:

A) Input attribute
B) Hidden attribute
C) Output attribute
D) Categorical attribute

Q54. Following are the types of supervised learning:

A) Regression
B) Classification
C) Subgroup discovery
D) All of above

Q55. A feature F1 can take certain values A–F, representing student grades. Feature type is:
A) Ordinal
B) Nominal
C) Categorical
D) Boolean

Q56. The output of training process in machine learning is:

A) Machine learning algorithm
B) Machine learning model
C) Null
D) Accuracy

Q57. Which of the following techniques would perform better for reducing dimensions of a
dataset?
A) Removing columns with high variance in data
B) Removing columns with too many missing values
C) Removing columns with dissimilar data trends
D) Removing columns with similar data trends
Q58. What characterizes a hyperplane in the geometrical model of machine learning?
A) A plane with 1 dimension fewer than number of input attributes
B) A plane with 1 more
C) A plane with 2 more
D) A plane with 2 fewer

Q59. Classifying reviews of a new Netflix series is an example of:

A) Unsupervised learning
B) Semi-supervised learning
C) Supervised learning
D) Reinforcement learning

Q60. Like the probabilistic view, the ___ view allows us to associate a probability of
membership with each classification:
A) Deductive
B) Exampler
C) Classical
D) Inductive

Q61. The problem of finding hidden structure in unlabeled data is called:

A) Unsupervised learning
B) Reinforcement learning
C) Supervised learning
D) Semi-supervised learning

Q62. Database query is used to uncover this type of knowledge:

A) Hidden
B) Shallow
C) Deep
D) Multidimensional

Q63. Data used to build a data mining model:

A) Training data
B) Hidden data
C) Test data
D) Validation data

Q64. Application of machine learning methods to large databases is called:

A) Big data computing
B) Artificial intelligence
C) Data mining
D) Internet of Things
Q65. Which would you address using a supervised learning algorithm?
A) Group web articles
B) Email labeled spam or not spam
C) Customer segmentation
D) Market basket patterns

Q66. In the stork-baby example, "number of babies" is:

A) Feature
B) Observation
C) Outcome
D) Attribute

Q67. A person trained to interact with a human expert to capture knowledge is called:
A) Knowledge developer
B) Knowledge programmer
C) Knowledge engineer
D) Knowledge extractor

Q68. What characterizes unlabeled examples in machine learning?

A) Plenty of confusing knowledge
B) Prior knowledge
C) No confusing knowledge
D) No prior knowledge

Q69. What does dimensionality reduction reduce?

A) Collinearity
B) Stochastic
C) Entropy
D) Performance

Q70. Segmenting telecom customers into groups is an example of:

A) Supervised learning
B) Unsupervised learning
C) Data extraction
D) Reinforcement learning

Q71. In multiclass classification, number of classes must be:

A) Equals to two
B) Less than two
C) Greater than two
D) Divisible by two

Q72. Which can only be used when training data are linearly separable?
A) Linear logistic regression
B) Linear hard-margin SVM
C) Linear soft-margin SVM
D) Parzen windows

Q73. Predicting earthquakes from seismic data is an example of:

A) Supervised learning
B) Unsupervised learning
C) Reinforcement learning
D) Dimensionality reduction

Q74. Prediction is:

A) A statistical projection
B) Value entered in database
C) Application of a rule to a specific case
D) Independent of data

Q75. Impact of high variance on the training set?

A) Underfitting
B) Overfitting
C) Both underfitting and overfitting
D) Depends on dataset

🔹 Q76–Q100 (DEC 2023 – 25 MCQs)

Q76. A feature F1 can take values A, B, C, D, E, & F and represents student grades. Feature type
is:
A) Ordinal
B) Nominal
C) Categorical
D) Boolean

Q77. Supervised learning and unsupervised clustering both require:

A) Input attribute
B) Hidden attribute
C) Output attribute
D) Categorical attribute

Q78. The output of training process in machine learning is:

A) Machine learning algorithm
B) Machine learning model
C) Null
D) Accuracy

Q79. What characterizes a hyperplane in the geometrical model of ML?

A) A plane with 1 dimension fewer than input attributes
B) A plane with 1 dimension more
C) A plane with 2 dimensions more
D) A plane with 2 dimensions fewer

Q80. Classifying Netflix reviews (positive/neutral/negative) is an example of:

A) Unsupervised learning
B) Semi-supervised learning
C) Supervised learning
D) Reinforcement learning

Q81. Like the probabilistic view, the ___ view allows associating probability of membership
with classification:
A) Deductive
B) Exampler
C) Classical
D) Inductive

Q82. Database query uncovers this type of knowledge:

A) Hidden
B) Shallow
C) Deep
D) Multidimensional

Q83. Data used to build a data mining model:

A) Training data
B) Hidden data
C) Test data
D) Validation data

Q84. Applying ML methods to large databases is known as:

A) Big data computing
B) Artificial intelligence
C) Data mining
D) Internet of Things

Q85. Which learning requires self-assessment to identify patterns?

A) Supervised learning
B) Unsupervised learning
C) Semi-supervised learning
D) Reinforced learning

Q86. Task to address using supervised learning:

A) Group news articles
B) Spam email filtering
C) Segment market
D) Market basket pattern

Q87. What type of learning uses labelled data?

A) Supervised learning
B) Unsupervised learning
C) Reinforcement learning
D) Active learning

Q88. In stork example, "number of babies" is:

A) Feature
B) Observation
C) Outcome
D) Attribute

Q89. Who interacts with human experts to capture knowledge?

A) Knowledge developer
B) Knowledge programmer
C) Knowledge engineer
D) Knowledge extractor

Q90. Unlabeled examples are characterized by:

A) Plenty of confusing knowledge
B) Prior knowledge
C) No confusing knowledge
D) No prior knowledge

Q91. Dimensionality reduction reduces:

A) Collinearity
B) Stochastic
C) Entropy
D) Performance

Q92. Segmenting telecom customers into groups is an example of:

A) Supervised learning
B) Unsupervised learning
C) Data extraction
D) Reinforcement learning

Q93. Which can only be used when training data are linearly separable?
A) Linear logistic regression
B) Linear soft-margin SVM
C) Linear hard-margin SVM
D) The centroid method
Q94. Predicting next earthquake from seismic data is:
A) Supervised learning
B) Unsupervised learning
C) Reinforcement learning
D) Dimensionality reduction

Q95. Prediction is:

A) Statistical projection
B) Expert-entered value
C) Result of applying a rule in a case
D) Independent of data

Q96. Impact of high variance on training set?

A) Underfitting
B) Overfitting
C) Both underfitting and overfitting
D) Depends on dataset

Q97. Example of feature extraction:

A) Applying PCA to project high dimensional data
B) Construction of bag-of-words
C) Removing stop words
D) Forward selection

Q98. How to choose principal components "k"?

A) Choose k = 99% of m
B) Choose smallest k retaining 99% variance
C) Choose largest k retaining 99% variance
D) Use elbow method

Q99. Student grade variable F1 taking A–D values is:

A) Ordinal variable
B) Not in mentioned categories
C) Nominal
D) Both ordinal and nominal

Q100. Which learning category does PCA fall under?

A) Classification
B) Feature extraction / dimensionality reduction
C) Reinforcement
D) Data labelling