Here is a 200-question (50 questions per chapter) objective-style exam

generated from your Table of Contents, with options A–D, complete with an
answer key.

The book title is COS 100: Artificial Intelligence.

---

COS 100: Artificial Intelligence

Multiple Choice Examination (200 Questions)

---

CHAPTER ONE: ARTIFICIAL INTELLIGENCE (Questions 1–50)

1. What does the preamble of Chapter One most likely introduce?

A) AI algorithms

B) Scope and background of AI

C) Python programming

D) Neural networks

✅ B) Scope and background of AI

2. Which of the following best defines Artificial Intelligence?

A) Simulating human intelligence in machines

B) Storing large amounts of data

C) Replacing all human jobs

D) Speeding up processors

✅ A) Simulating human intelligence in machines

3. The idea of AI originated in which decade?

A) 1940s

B) 1950s

C) 1960s

D) 1980s

✅ B) 1950s

4. Who is credited with organizing the 1956 Dartmouth Workshop?

A) Alan Turing

B) John McCarthy

C) Marvin Minsky

D) Geoffrey Hinton

✅ B) John McCarthy

5. Which era saw the development of expert systems?

A) 1950s

B) 1960s

C) 1970s–80s

D) 1990s

✅ C) 1970s–80s

6. Narrow AI refers to systems that:

A) Perform any intellectual task like a human

B) Excel in a single task

C) Have self-awareness

D) Use only symbolic reasoning

✅ B) Excel in a single task

7. Which is an example of General AI?

A) Chess-playing program

B) Self-driving car

C) Human-like reasoning across domains (not yet achieved)

D) Chatbot for customer service

✅ C) Human-like reasoning across domains (not yet achieved)

8. Superintelligence refers to AI that:

A) Works only in games

B) Surpasses human intelligence in all fields

C) Requires constant supervision

D) Is limited to narrow tasks

✅ B) Surpasses human intelligence in all fields

9. Reactive machines in AI:

A) Have memory of past actions

B) Cannot use past experiences

C) Are self-aware

D) Use deep learning always

✅ B) Cannot use past experiences

10. Which type of AI includes theory of mind?

A) Reactive machines
B) Limited memory

C) Theory of mind AI

D) Superintelligence

✅ C) Theory of mind AI

11. Importance of AI in modern society includes:

A) Automating routine tasks

B) Improving healthcare diagnostics

C) Enhancing customer service

D) All of the above

✅ D) All of the above

12. AI in agriculture helps with:

A) Crop monitoring

B) Soil analysis

C) Pest detection

D) All of the above

✅ D) All of the above

13. Which sub-field of AI focuses on enabling machines to see?

A) NLP

B) Computer Vision

C) Expert Systems

D) Robotics

✅ B) Computer Vision

14. Natural Language Processing (NLP) deals with:

A) Image recognition

B) Speech and text understanding

C) Robot movement

D) Database queries

✅ B) Speech and text understanding

15. Expert systems emulate:

A) Human emotions

B) Human decision-making in specific domains

C) General intelligence

D) Neural networks only

✅ B) Human decision-making in specific domains

16. Which is an AI application in finance?

A) Fraud detection

B) Algorithmic trading

C) Credit scoring

D) All of the above

✅ D) All of above

17. AI in entertainment includes:

A) Recommendation systems

B) Game AI

C) Content generation

D) All of the above

✅ D) All of the above

18. Sub-fields of AI include:

A) Machine Learning

B) Robotics

C) Expert Systems

D) All of the above

✅ D) All of the above

19. Which type of AI is currently most common?

A) General AI

B) Superintelligence

C) Narrow AI

D) Conscious AI

✅ C) Narrow AI

20. AI winter refers to:

A) Cold weather effects on computers

B) Period of reduced funding and interest

C) AI taking holidays

D) Seasonal AI updates

✅ B) Period of reduced funding and interest

21. The Turing Test measures:

A) Processing speed

B) Ability to exhibit human-like conversation

C) Memory capacity

D) Energy efficiency

✅ B) Ability to exhibit human-like conversation

22. Which of the following is NOT a kind of AI by capability?

A) Reactive machines

B) Limited memory

C) Quantum AI

D) Self-aware AI

✅ C) Quantum AI

23. Self-aware AI is:

A) Fully operational today

B) Hypothetical future AI

C) Used in spam filters

D) A type of narrow AI

✅ B) Hypothetical future AI

24. AI in healthcare includes:

A) Medical imaging analysis

B) Drug discovery

C) Patient monitoring

D) All of the above

✅ D) All of the above

25. Which language is commonly used in AI development?

A) HTML

B) Python

C) CSS

D) Assembly
✅ B) Python

26. The term "Artificial Intelligence" was coined in:

A) 1943

B) 1956

C) 1969

D) 1980

✅ B) 1956

27. Deep Blue, which beat Kasparov, is an example of:

A) General AI

B) Narrow AI

C) Superintelligence

D) Self-aware AI

✅ B) Narrow AI

28. AI in transportation includes:

A) Autonomous vehicles

B) Traffic prediction

C) Route optimization

D) All of the above

✅ D) All of the above

29. AI ethics deals with:

A) Bias in algorithms

B) Privacy concerns

C) Job displacement
D) All of the above

✅ D) All of the above

30. Which is NOT a sub-field of AI?

A) Machine Learning

B) Cryptography

C) Robotics

D) NLP

✅ B) Cryptography

31. AI in cybersecurity helps by:

A) Detecting intrusions

B) Automating responses

C) Analyzing malware

D) All of the above

✅ D) All of the above

32. The Chinese Room argument challenges:

A) AI processing speed

B) Strong AI claims about understanding

C) Machine learning accuracy

D) Robot navigation

✅ B) Strong AI claims about understanding

33. Which sector uses AI for predictive maintenance?

A) Manufacturing

B) Retail
C) Banking

D) Education

✅ A) Manufacturing

34. AI in education includes:

A) Personalized tutoring

B) Automated grading

C) Student analytics

D) All of the above

✅ D) All of the above

35. Which of the following is a kind of AI by functionality?

A) Reactive machines

B) Symbolic AI

C) Connectionist AI

D) Hybrid AI

✅ A) Reactive machines

36. History of AI includes the term "Perceptron" developed by:

A) Rosenblatt

B) Turing

C) McCarthy

D) Minsky

✅ A) Rosenblatt

37. AI in robotics helps with:

A) Motion planning
B) Perception

C) Control

D) All of the above

✅ D) All of the above

38. Which AI sub-field deals with human-like reasoning?

A) Knowledge representation

B) Computer vision

C) Speech recognition

D) Big data

✅ A) Knowledge representation

39. The Dartmouth workshop is significant because it:

A) Created the first robot

B) Established AI as a field

C) Invented the internet

D) Developed ChatGPT

✅ B) Established AI as a field

40. AI in law includes:

A) Document review

B) Legal research

C) Predictive justice

D) All of the above

✅ D) All of the above

41. Which is NOT a benefit of AI?

A) 24/7 availability

B) Zero bias always

C) Handling large data

D) Automation of repetitive tasks

✅ B) Zero bias always

42. Which type of AI can remember past decisions?

A) Reactive machines

B) Limited memory

C) Theory of mind

D) Self-aware

✅ B) Limited memory

43. AI in gaming is used for:

A) Non-player characters

B) Difficulty adjustment

C) Procedural content

D) All of the above

✅ D) All of the above

44. The LOEBner Prize is associated with:

A) Chess AI

B) Turing Test competitions

C) Robot soccer

D) Data mining

✅ B) Turing Test competitions

45. AI winter was caused partly by:

A) Too much success

B) Overpromising and underdelivering

C) Lack of computers

D) Government bans

✅ B) Overpromising and underdelivering

46. Which of the following is a modern AI application?

A) Virtual assistants (Siri, Alexa)

B) Email spam filters

C) Recommendation engines

D) All of the above

✅ D) All of the above

47. The concept of "neural networks" is inspired by:

A) Human brain

B) Ant colonies

C) Solar systems

D) Cloud computing

✅ A) Human brain

48. Which of the following is a challenge in AI?

A) Explainability

B) Data privacy

C) Computational cost

D) All of the above

✅ D) All of the above

49. AI in social media includes:

A) Content moderation

B) Ad targeting

C) Friend suggestions

D) All of the above

✅ D) All of the above

50. Which of the following statements is true?

A) AI has no limitations

B) AI can completely replace human creativity

C) AI systems can exhibit bias

D) AI never needs data

✅ C) AI systems can exhibit bias

---

CHAPTER TWO: INTELLIGENT SYSTEMS AND AGENTS (Questions 51–100)

51. An intelligent system can:

A) Only perform calculations

B) Perceive its environment and act

C) Never learn

D) Work without data

✅ B) Perceive its environment and act

52. Properties of intelligent systems include:

A) Adaptability

B) Learning ability

C) Reactivity

D) All of the above

✅ D) All of the above

53. An intelligent agent is anything that:

A) Stores data

B) Acts autonomously to achieve goals

C) Has a human form

D) Runs on Windows

✅ B) Acts autonomously to achieve goals

54. The agent-environment interaction model includes:

A) Perception, action, feedback

B) Only output

C) No feedback loop

D) Human intervention only

✅ A) Perception, action, feedback

55. A thermostat is an example of:

A) General AI

B) Simple reflex agent

C) Superintelligence

D) NLP system

✅ B) Simple reflex agent

56. Which is a real-world intelligent system?

A) Roomba vacuum cleaner

B) Stone

C) Manual typewriter

D) Flashlight

✅ A) Roomba vacuum cleaner

57. Model-based reflex agents:

A) Have no memory

B) Maintain internal state

C) Ignore history

D) Are purely reactive

✅ B) Maintain internal state

58. Goal-based agents:

A) Act randomly

B) Consider future consequences

C) Never plan

D) Only react

✅ B) Consider future consequences

59. Utility-based agents aim to:

A) Satisfice only

B) Maximize a performance measure

C) Minimize effort

D) Ignore goals

✅ B) Maximize a performance measure

60. Learning agents have which component?

A) Critic

B) Performance element

C) Learning element

D) All of the above

✅ D) All of the above

61. A chess-playing AI is a:

A) Simple reflex agent

B) Goal-based agent

C) Reactive agent

D) Non-learning agent only

✅ B) Goal-based agent

62. The environment in AI agents is:

A) Always fully observable

B) Everything outside the agent

C) Same as the agent

D) Virtual only

✅ B) Everything outside the agent

63. A deterministic environment means:

A) Next state is fully determined by current state and action

B) Random outcomes

C) No actions possible

D) Unpredictable
✅ A) Next state is fully determined by current state and action

64. In a partially observable environment:

A) Agent sees all

B) Agent lacks full state information

C) No sensors exist

D) Only deterministic

✅ B) Agent lacks full state information

65. A single-agent environment:

A) Has one agent acting

B) Requires multiple agents

C) Is always competitive

D) No interaction

✅ A) Has one agent acting

66. Multi-agent environments can be:

A) Cooperative

B) Competitive

C) Mixed

D) All of the above

✅ D) All of the above

67. An intelligent system differs from an intelligent agent in that:

A) System is broader, agents are components

B) Agent is broader

C) No difference
D) System cannot learn

✅ A) System is broader, agents are components

68. Which agent type uses condition-action rules?

A) Simple reflex

B) Learning

C) Utility-based

D) Goal-based

✅ A) Simple reflex

69. Rational agent definition:

A) Always succeeds

B) Does the right thing based on knowledge

C) Never makes errors

D) Needs no sensors

✅ B) Does the right thing based on knowledge

70. Performance measure for an agent:

A) Speed only

B) Criteria for success

C) Memory usage

D) Code length

✅ B) Criteria for success

71. An example of a dynamic environment:

A) Chess with timed moves

B) Static puzzle
C) Photo

D) Book

✅ A) Chess with timed moves

72. An environment is episodic if:

A) Each action depends on past

B) Each episode is independent

C) No actions possible

D) Continuous actions

✅ B) Each episode is independent

73. Intelligent agents use sensors to:

A) Act

B) Perceive

C) Sleep

D) Compute

✅ B) Perceive

74. Actuators in agents are used for:

A) Thinking

B) Acting on environment

C) Storing data

D) Learning

✅ B) Acting on environment

75. A robot vacuum cleaner is which type of system?

A) Intelligent system
B) Simple reflex

C) Learning agent possible

D) All of the above

✅ D) All of the above

76. Categorization of AI agents includes:

A) Reflex, model-based, goal-based, utility-based, learning

B) Only learning agents

C) Only reactive

D) No categories

✅ A) Reflex, model-based, goal-based, utility-based, learning

77. Use case for learning agent:

A) Spam filter adapting to new emails

B) Calculator

C) Light switch

D) Bicycle

✅ A) Spam filter adapting to new emails

78. Use case for utility-based agent:

A) Self-driving car optimizing safety and speed

B) Clock

C) Desk lamp

D) Hammer

✅ A) Self-driving car optimizing safety and speed

79. Which is NOT a property of intelligent systems?

A) Reactivity

B) Proactiveness

C) Inability to learn

D) Social ability

✅ C) Inability to learn

80. Software agents are:

A) Physical robots

B) Programs acting autonomously

C) Hardware only

D) Never intelligent

✅ B) Programs acting autonomously

81. A stock trading bot is an example of:

A) Physical agent

B) Software agent

C) Simple reflex

D) Non-learning

✅ B) Software agent

82. In agent-environment model, perception leads to:

A) Action

B) Inaction

C) Deletion

D) Error

✅ A) Action
83. A fully observable environment:

A) Agent sees all relevant aspects

B) Hidden information exists

C) No sensors

D) Random

✅ A) Agent sees all relevant aspects

84. Discrete environment example:

A) Chess board states

B) Continuous temperature

C) Rainfall amount

D) Speed of car

✅ A) Chess board states

85. Continuous environment example:

A) Real-time driving

B) Tic-tac-toe

C) Puzzle solving

D) Turn-based game

✅ A) Real-time driving

86. An intelligent system must be:

A) Autonomous at least partially

B) Always correct

C) Humanoid

D) Slow

✅ A) Autonomous at least partially

87. Which agent has internal model of world?

A) Model-based reflex

B) Simple reflex

C) Random agent

D) Null agent

✅ A) Model-based reflex

88. Goal-based agent uses:

A) Search and planning

B) No memory

C) Random actions

D) Fixed rules only

✅ A) Search and planning

89. Utility function helps compare:

A) Different outcomes

B) CPU speed

C) Memory size

D) Color

✅ A) Different outcomes

90. Learning agent's performance element:

A) Selects actions

B) Learns new strategies

C) Stores all data forever

D) Ignores feedback
✅ A) Selects actions

91. Critic in learning agent:

A) Gives feedback on performance

B) Punishes the agent

C) Deletes old knowledge

D) Disables learning

✅ A) Gives feedback on performance

92. Which environment is most challenging for AI?

A) Fully observable, deterministic

B) Partially observable, stochastic

C) Episodic, static

D) Single-agent

✅ B) Partially observable, stochastic

93. Email spam filter agent is which type?

A) Simple reflex

B) Learning agent

C) Only rule-based

D) Utility-based without learning

✅ B) Learning agent

94. Intelligent systems can be found in:

A) Smart homes

B) Medical diagnosis

C) Autonomous drones
D) All of the above

✅ D) All of the above

95. Which is a use case for multi-agent systems?

A) Traffic control simulation

B) Single-player game

C) Static website

D) Paper notebook

✅ A) Traffic control simulation

96. Proactiveness means agent:

A) Reacts only

B) Takes initiative

C) Never acts

D) Needs human start

✅ B) Takes initiative

97. Social ability in agents means:

A) Interacting with other agents/humans

B) Being friendly

C) Having emotions

D) Using social media

✅ A) Interacting with other agents/humans

98. A rational agent chooses actions to:

A) Minimize performance measure

B) Maximize expected performance

C) Be random

D) Avoid all risks

✅ B) Maximize expected performance

99. Which best describes an intelligent agent?

A) Always physical

B) Autonomous, reactive, goal-directed

C) Cannot learn

D) Needs Internet always

✅ B) Autonomous, reactive, goal-directed

100. The main difference between intelligent systems and agents is:

A) Systems are hardware, agents are software

B) Systems may include multiple agents

C) No difference

D) Agents cannot be intelligent

✅ B) Systems may include multiple agents

---

CHAPTER THREE: PROBLEM SOLVING AND SEARCH TECHNIQUES (Questions

101–150)

101. Problem solving in AI typically involves:

A) Searching for a solution in state space

B) Random guessing

C) No formal method
D) Only mathematical formulas

✅ A) Searching for a solution in state space

102. Problem representation in AI includes:

A) States, actions, goal test, path cost

B) Only initial state

C) No goal

D) Irrelevant data

✅ A) States, actions, goal test, path cost

103. State space representation is:

A) Set of all possible states

B) Only final states

C) Only initial state

D) Actions without states

✅ A) Set of all possible states

104. A node in state space represents:

A) An action

B) A state

C) A cost

D) A solution

✅ B) A state

105. An edge in state space represents:

A) Another state

B) An action
C) Cost only

D) Goal

✅ B) An action

106. The initial state is:

A) Where the agent starts

B) Final state

C) Random state

D) No state

✅ A) Where the agent starts

107. Goal test determines:

A) If current state is goal

B) Next action

C) Path cost

D) Heuristic value

✅ A) If current state is goal

108. Path cost function:

A) Assigns cost to each path

B) Ignores steps

C) Only for final state

D) Random

✅ A) Assigns cost to each path

109. Uninformed search uses:

A) Domain knowledge
B) No additional info beyond problem definition

C) Heuristics

D) Machine learning

✅ B) No additional info beyond problem definition

110. Informed search uses:

A) Heuristics

B) No knowledge

C) Only BFS

D) Only DFS

✅ A) Heuristics

111. Breadth-first search (BFS) is:

A) Uninformed

B) Informed

C) Heuristic

D) Greedy

✅ A) Uninformed

112. Depth-first search (DFS) uses:

A) LIFO stack

B) FIFO queue

C) Priority queue

D) Random order

✅ A) LIFO stack

113. BFS guarantees:

A) Shortest path in unweighted graphs

B) Always memory efficient

C) Uses heuristics

D) Depth-first order

✅ A) Shortest path in unweighted graphs

114. DFS is better than BFS when:

A) Memory is limited

B) Shortest path needed

C) All solutions shallow

D) Graph is small

✅ A) Memory is limited

115. A search uses:*

A) f(n) = g(n) + h(n)

B) f(n) = g(n) only

C) f(n) = h(n) only

D) Random function

✅ A) f(n) = g(n) + h(n)

116. A heuristic is admissible if:

A) It overestimates cost

B) It never overestimates true cost

C) It is random

D) It equals zero

✅ B) It never overestimates true cost

117. Greedy best-first search uses:

A) f(n) = h(n)

B) f(n) = g(n)

C) f(n) = g(n)+h(n)

D) No function

✅ A) f(n) = h(n)

118. Which algorithm is complete but not optimal?

A) DFS

B) BFS

C) A* with admissible heuristic

D) Uniform cost search

✅ A) DFS (if infinite depth, not complete; but in finite tree, complete, not
optimal)

119. Uniform cost search expands:

A) Node with lowest path cost

B) Node with highest heuristic

C) Random node

D) Deepest node

✅ A) Node with lowest path cost

120. Bidirectional search requires:

A) Two searches meeting

B) Only forward search

C) Only backward

D) No goal
✅ A) Two searches meeting

121. State space explosion refers to:

A) Too many states

B) No states

C) Easy problems

D) Small memory

✅ A) Too many states

122. Heuristic functions are used to:

A) Guide search efficiently

B) Randomize search

C) Slow down search

D) Remove states

✅ A) Guide search efficiently

123. The 8-puzzle is a classic problem for:

A) Search algorithms

B) Sorting

C) Encryption

D) Graphics

✅ A) Search algorithms

124. Which is uninformed search?

A) Iterative deepening DFS

B) A*

C) Greedy BFS
D) Hill climbing

✅ A) Iterative deepening DFS

125. Iterative deepening DFS combines:

A) BFS memory and DFS completeness

B) Only BFS

C) Only DFS

D) Random search

✅ A) BFS memory and DFS completeness

126. Hill climbing is prone to:

A) Local maxima

B) Global optimum always

C) No plateaus

D) Fast guaranteed optimum

✅ A) Local maxima

127. Simulated annealing allows:

A) Sometimes worse moves to escape local optima

B) Only better moves

C) No randomness

D) Only deterministic

✅ A) Sometimes worse moves to escape local optima

128. Genetic algorithms are inspired by:

A) Evolution

B) Thermodynamics
C) Quantum physics

D) Calculus

✅ A) Evolution

129. The traveling salesman problem is solved by:

A) Search techniques

B) Only brute force

C) No algorithm exists

D) Sorting

✅ A) Search techniques

130. Which is optimal and complete with admissible heuristic?

A) A*

B) Greedy BFS

C) Hill climbing

D) DFS

✅ A) A*

131. In problem solving, a solution is:

A) Path from initial to goal

B) Any state

C) Only final action

D) Random node

✅ A) Path from initial to goal

132. Search tree vs state space graph:

A) Tree may have repeated states

B) Same thing

C) Graph has no cycles

D) Tree always smaller

✅ A) Tree may have repeated states

133. Graph search avoids loops by:

A) Keeping explored set

B) Adding more states

C) Ignoring cycles

D) BFS only

✅ A) Keeping explored set

134. Real-time search requires:

A) Time-limited decisions

B) Unlimited time

C) No constraints

D) Only offline

✅ A) Time-limited decisions

135. Which search uses priority queue?

A) Uniform cost

B) DFS

C) BFS without queue

D) Random

✅ A) Uniform cost

136. Depth-limited search solves:

A) Infinite state space

B) Only small problems

C) No goal

D) Uninformed only

✅ A) Infinite state space

137. The missionaries and cannibals problem tests:

A) Constraint satisfaction

B) Sorting

C) Compression

D) Encryption

✅ A) Constraint satisfaction

138. Constraint satisfaction problems (CSP):

A) Variables, domains, constraints

B) No variables

C) Only one solution

D) No constraints

✅ A) Variables, domains, constraints

139. Backtracking is used in:

A) CSP solving

B) Sorting arrays

C) Graphics rendering

D) Memory allocation

✅ A) CSP solving
140. Forward checking in CSP:

A) Looks ahead to prune search

B) Checks only after assignment

C) No pruning

D) Random

✅ A) Looks ahead to prune search

141. Which is a local search algorithm?

A) Hill climbing

B) BFS

C) A*

D) Uniform cost

✅ A) Hill climbing

142. Beam search keeps:

A) k best nodes

B) only best node

C) all nodes

D) random nodes

✅ A) k best nodes

143. Bidirectional search is efficient when:

A) Branching factor same in both directions

B) Only forward possible

C) Goal unknown

D) No initial state

✅ A) Branching factor same in both directions

144. Which is NOT a search technique?

A) Heuristic evaluation

B) Compilation

C) Backtracking

D) Branch and bound

✅ B) Compilation

145. Branch and bound uses:

A) Pruning based on cost bound

B) No pruning

C) Only DFS

D) Only BFS

✅ A) Pruning based on cost bound

146. The water jug problem is solved by:

A) State space search

B) Neural networks only

C) Clustering

D) Regression

✅ A) State space search

147. Heuristic function h(n) estimates:

A) Cost to goal from n

B) Cost from start to n

C) Total cost

D) Random value
✅ A) Cost to goal from n

148. Manhattan distance is a heuristic for:

A) Grid problems

B) Continuous problems

C) Sorting

D) Compression

✅ A) Grid problems

149. Which algorithm is optimal for uniform cost?

A) Uniform cost search

B) DFS

C) Hill climbing

D) Greedy BFS

✅ A) Uniform cost search

150. Problem solving in AI ends when:

A) Goal state reached

B) No memory left

C) Time runs out

D) Random stop

✅ A) Goal state reached

---

CHAPTER FOUR: INTRODUCTION TO MACHINE LEARNING (Questions 151–

200)
151. Machine Learning is a subset of:

A) Artificial Intelligence

B) Deep Learning

C) Statistics only

D) Robotics

✅ A) Artificial Intelligence

152. Deep Learning is a subset of:

A) Machine Learning

B) Expert Systems

C) NLP only

D) Symbolic AI

✅ A) Machine Learning

153. Which is true about ML?

A) ML systems learn from data

B) ML never improves with experience

C) ML requires explicit programming for all tasks

D) ML is not part of AI

✅ A) ML systems learn from data

154. Supervised learning uses:

A) Labeled data

B) Unlabeled data

C) No data

D) Only reinforcement
✅ A) Labeled data

155. Unsupervised learning finds:

A) Patterns without labels

B) Only classification

C) Only regression

D) Rewards

✅ A) Patterns without labels

156. Reinforcement learning uses:

A) Rewards and punishments

B) Labeled examples only

C) No feedback

D) Only unsupervised

✅ A) Rewards and punishments

157. Classification is a type of:

A) Supervised learning

B) Unsupervised learning

C) Reinforcement learning

D) Clustering only

✅ A) Supervised learning

158. Regression predicts:

A) Continuous values

B) Categories

C) Clusters
D) Rewards

✅ A) Continuous values

159. Clustering is an example of:

A) Unsupervised learning

B) Supervised learning

C) Reinforcement learning

D) Deep learning only

✅ A) Unsupervised learning

160. The ML workflow includes:

A) Data collection, preprocessing, training, evaluation

B) Only model training

C) Only data collection

D) No evaluation

✅ A) Data collection, preprocessing, training, evaluation

161. Overfitting occurs when:

A) Model learns noise too well

B) Model underperforms

C) No training

D) Simple model

✅ A) Model learns noise too well

162. Underfitting means:

A) Model too simple to capture patterns

B) Model too complex

C) Perfect generalization

D) No data

✅ A) Model too simple to capture patterns

163. Training set is used to:

A) Learn model parameters

B) Test final accuracy

C) Validate only

D) Never used

✅ A) Learn model parameters

164. Test set is used to:

A) Evaluate final model

B) Train the model

C) Overfit

D) Increase bias

✅ A) Evaluate final model

165. Validation set helps:

A) Tune hyperparameters

B) Train initial model

C) Forget data

D) Overfit more

✅ A) Tune hyperparameters

166. Which is NOT a type of ML?

A) Supervised
B) Unsupervised

C) Transductive

D) Reinforcement

✅ C) Transductive (not standard main type, but semi-supervised exists;

standard three: supervised, unsupervised, reinforcement)

167. k-Nearest Neighbors is:

A) Supervised learning

B) Unsupervised

C) Reinforcement

D) Clustering only

✅ A) Supervised learning

168. k-Means is:

A) Unsupervised clustering

B) Supervised regression

C) Classification

D) Reinforcement

✅ A) Unsupervised clustering

169. Linear regression is used for:

A) Predicting continuous values

B) Classification

C) Clustering

D) Association rules

✅ A) Predicting continuous values

170. Decision trees can be used for:

A) Both classification and regression

B) Only clustering

C) Only reinforcement

D) Only NLP

✅ A) Both classification and regression

171. Neural networks are inspired by:

A) Biological neurons

B) Statistical tests

C) Decision trees

D) Sorting algorithms

✅ A) Biological neurons

172. Deep learning uses:

A) Many hidden layers

B) Only one layer

C) No layers

D) Only linear functions

✅ A) Many hidden layers

173. Which is true about AI, ML, DL?

A) DL ⊂ ML ⊂ AI

B) AI ⊂ ML ⊂ DL

C) ML ⊂ AI ⊂ DL

D) They are unrelated

✅ A) DL ⊂ ML ⊂ AI
174. Feature extraction in ML means:

A) Selecting relevant input variables

B) Adding noise

C) Removing all data

D) Increasing overfitting

✅ A) Selecting relevant input variables

175. Bias-variance tradeoff:

A) Balancing underfitting and overfitting

B) Always high bias

C) Always high variance

D) No relation

✅ A) Balancing underfitting and overfitting

176. A confusion matrix is used for:

A) Classification evaluation

B) Regression

C) Clustering

D) Reinforcement

✅ A) Classification evaluation

177. Precision measures:

A) True positives / (True positives + False positives)

B) True positives / Actual positives

C) Overall accuracy

D) False negatives
✅ A) True positives / (True positives + False positives)

178. Recall measures:

A) True positives / (True positives + False negatives)

B) True positives / Predicted positives

C) False positives

D) Accuracy

✅ A) True positives / (True positives + False negatives)

179. F1-score is:

A) Harmonic mean of precision and recall

B) Arithmetic mean

C) Geometric mean

D) Sum of precision and recall

✅ A) Harmonic mean of precision and recall

180. ROC curve plots:

A) TPR vs FPR

B) Precision vs recall

C) Accuracy vs loss

D) Bias vs variance

✅ A) TPR vs FPR

181. Which algorithm is used for dimensionality reduction?

A) PCA

B) SVM

C) KNN
D) Linear regression

✅ A) PCA

182. Gradient descent optimizes:

A) Loss function

B) Data size

C) Number of features

D) Input noise

✅ A) Loss function

183. Learning rate controls:

A) Step size in gradient descent

B) Number of epochs

C) Training data

D) Model depth

✅ A) Step size in gradient descent

184. Epoch in ML means:

A) One pass through entire training data

B) One batch

C) One sample

D) One parameter update

✅ A) One pass through entire training data

185. Batch size is:

A) Number of samples per gradient update

B) Total dataset size

C) Number of epochs

D) Model size

✅ A) Number of samples per gradient update

186. Overfitting can be reduced by:

A) Regularization

B) Increasing model complexity

C) Removing training data

D) Ignoring validation

✅ A) Regularization

187. Dropout is used in:

A) Neural networks to prevent overfitting

B) Linear regression

C) Decision trees

D) k-Means

✅ A) Neural networks to prevent overfitting

188. Cross-validation helps:

A) Assess model generalization

B) Increase overfitting

C) Reduce data

D) Skip testing

✅ A) Assess model generalization

189. Reinforcement learning agent learns:

A) Policy mapping states to actions

B) Only from labeled data

C) Without rewards

D) Clusters

✅ A) Policy mapping states to actions

190. Q-learning is a:

A) Reinforcement learning algorithm

B) Supervised algorithm

C) Unsupervised algorithm

D) Clustering method

✅ A) Reinforcement learning algorithm

191. Which is an application of ML?

A) Spam detection

B) Face recognition

C) Recommendation systems

D) All of the above

✅ D) All of the above

192. The "shot on AWESOME A70" in your TOC likely indicates:

A) A placeholder or note

B) An ML algorithm

C) A dataset

D) A loss function

✅ A) A placeholder or note

193. Semi-supervised learning uses:

A) Small labeled + large unlabeled data

B) Only labeled

C) Only unlabeled

D) No data

✅ A) Small labeled + large unlabeled data

194. Transfer learning:

A) Reusing a pre-trained model on new task

B) Learning from scratch

C) No reuse

D) Only for vision

✅ A) Reusing a pre-trained model on new task

195. Ensemble methods combine:

A) Multiple models

B) Single model only

C) Only data

D) Only loss functions

✅ A) Multiple models

196. Random Forest is:

A) Ensemble of decision trees

B) Single tree

C) Neural network

D) Clustering

✅ A) Ensemble of decision trees

197. Which is a challenge in ML?

A) Data quality

B) Computational cost

C) Interpretability

D) All of the above

✅ D) All of the above

198. Online learning updates model:

A) Incrementally as data arrives

B) Only once

C) Batch only

D) Never

✅ A) Incrementally as data arrives

199. The "No Free Lunch" theorem in ML states:

A) No single algorithm works best for all problems

B) Free lunch provided

C) All algorithms equal

D) No algorithms work

✅ A) No single algorithm works best for all problems

200. Which is NOT part of typical ML workflow?

A) Compiling source code

B) Data preprocessing

C) Model training

D) Evaluation

✅ A) Compiling source code

---

ANSWER KEY (Chapters 1–4)

Q# Answer Q# Answer Q# Answer Q# Answer

1 B 51 B 101 A 151 A

2 A 52 D 102 A 152 A

3 B 53 B 103 A 153 A

4 B 54 A 104 B 154 A

5 C 55 B 105 B 155 A

6 B 56 A 106 A 156 A

7 C 57 B 107 A 157 A

8 B 58 B 108 A 158 A

9 B 59 B 109 B 159 A

10 C 60 D 110 A 160 A

11 D 61 B 111 A 161 A

12 D 62 B 112 A 162 A

13 B 63 A 113 A 163 A

14 B 64 B 114 A 164 A

15 B 65 A 115 A 165 A

16 D 66 D 116 B 166 C

17 D 67 A 117 A 167 A

18 D 68 A 118 A 168 A

19 C 69 B 119 A 169 A

20 B 70 B 120 A 170 A

21 B 71 A 121 A 171 A
22 C 72 B 122 A 172 A

23 B 73 B 123 A 173 A

24 D 74 B 124 A 174 A

25 B 75 D 125 A 175 A

26 B 76 A 126 A 176 A

27 B 77 A 127 A 177 A

28 D 78 A 128 A 178 A

29 D 79 C 129 A 179 A

30 B 80 B 130 A 180 A

31 D 81 B 131 A 181 A

32 B 82 A 132 A 182 A

33 A 83 A 133 A 183 A

34 D 84 A 134 A 184 A

35 A 85 A 135 A 185 A

36 A 86 A 136 A 186 A

37 D 87 A 137 A 187 A

38 A 88 A 138 A 188 A

39 B 89 A 139 A 189 A

40 D 90 A 140 A 190 A

41 B 91 A 141 A 191 D

42 B 92 B 142 A 192 A

43 D 93 B 143 A 193 A

44 B 94 D 144 B 194 A

45 B 95 A 145 A 195 A

46 D 96 B 146 A 196 A

47 A 97 A 147 A 197 D

48 D 98 B 148 A 198 A
49 D 99 B 149 A 199 A

50 C 100 B 150 A 200 A

---

Let me know if you want this exported as a PDF file or turned into a printable
exam booklet with title page and instructions.