Set 1
Introduction to Artificial Intelligence
1. Explain the core problems in Artificial Intelligence (AI). How do AI techniques address
these problems?
2. Discuss the Tic-Tac-Toe problem as an example of state-space representation in AI.
Intelligent Agents
3. Differentiate between the types of agents: goal-based, utility-based, and learning agents,
providing examples for each.
4. How does the nature of the environment (deterministic vs. stochastic) influence the
design of an intelligent agent?
Problem Solving and Search
5. Define a problem as a state-space search. What are the key components involved in this
representation?
6. Compare breadth-first and depth-first search strategies in terms of time complexity,
space complexity, and completeness.
7. Explain the concept of bidirectional search. Under what conditions is it more efficient
than other search strategies?
Heuristic and Optimization Search
8. Derive the evaluation function used in the A* search algorithm. Why is it considered
optimal and complete?
9. Discuss the limitations of hill climbing search. How does simulated annealing overcome
these limitations?
10. Provide a detailed explanation of genetic algorithms as an optimization technique. How
does it emulate biological evolution?
Constraint Satisfaction Problems (CSP)
11. Solve a given CSP using backtracking and constraint propagation. Compare the
efficiency of the two approaches.
12. What are local search algorithms for CSP? Discuss their applications and limitations.
Adversarial Search
13. Describe the minimax search procedure and illustrate it with an example involving a two-
player game.
14. Explain alpha-beta pruning and its significance in reducing the complexity of minimax
search.
15. Discuss the concept of iterative deepening in adversarial search and its advantages in
real-time systems.
Knowledge and Reasoning
16. Identify and analyze the issues in knowledge representation. How do they affect the
performance of AI systems?
17. Discuss the differences between logical and probabilistic approaches to knowledge
representation.
Using Predicate Logic
18. Represent the following in predicate logic:
● "All humans are mortal."
● "Socrates is a human."
● "Therefore, Socrates is mortal."
19. Explain the resolution method in predicate logic with an example.
Probabilistic Reasoning
20. Define Bayesian networks. How do they handle uncertainty in AI systems?
21. Compare and contrast Bayesian reasoning, Dempster-Shafer theory, and fuzzy logic in
uncertain domains.
Natural Language Processing (NLP)
22. Describe the steps involved in syntactic and semantic analysis in natural language
processing.
23. Discuss the challenges of discourse and pragmatic processing in NLP systems.
Learning in AI
24. Explain the differences between inductive learning and explanation-based learning with
examples.
25. Illustrate the process of decision tree learning with a real-world dataset.
26. Discuss the advantages of neural network learning over traditional machine learning
techniques.
Expert Systems
27. What are expert systems, and how do they represent and use domain knowledge?
Provide an example.
28. Explain the structure and functionality of an expert system shell.
29. Discuss knowledge acquisition techniques in the development of expert systems.
Set 2:
Introduction to Artificial Intelligence
1. What is the fundamental difference between an AI technique and a conventional
algorithm?
2. Why is the Tic-Tac-Toe problem often used as a starting point for AI problem-solving
concepts?
Intelligent Agents
3. How does the structure of an intelligent agent determine its behavior in an environment?
4. Explain the trade-offs between utility-based agents and goal-based agents in decision-
making.
Problem Solving and Search
5. Why is defining a problem as a state-space search important for AI systems?
6. Compare the benefits and limitations of systematic search strategies (e.g., BFS and
DFS) versus heuristic-based searches (e.g., A*).
Heuristic and Optimization Search
7. What makes A* search optimal and complete, and why is this property significant for AI
applications?
8. How does simulated annealing draw parallels to the physical process of annealing in
metallurgy?
9. Why are genetic algorithms considered robust for solving complex optimization
problems?
Constraint Satisfaction Problems (CSP)
10. What is the significance of constraint propagation in solving CSPs efficiently?
11. Why are local search techniques effective for some CSPs but not for others?
Adversarial Search
12. What are the primary challenges in designing algorithms for adversarial search in games
like chess or Go? [out of syllabus but you can try]
13. How does alpha-beta pruning improve the efficiency of the minimax algorithm, and why
is it critical for real-time decision-making?
Knowledge and Reasoning
14. Why is knowledge representation a critical component of AI systems?
15. What are the main challenges in mapping real-world problems into formal knowledge
representation schemes?
Using Predicate Logic
16. How does predicate logic enable the representation of complex relationships in AI?
17. What are the advantages and limitations of using resolution for logical inference?
Probabilistic Reasoning
18. Why is probabilistic reasoning essential for AI systems operating in uncertain
environments?
19. Compare the approaches of Bayesian networks and fuzzy logic for handling uncertainty.
Which would you choose for a weather prediction system and why?
Natural Language Processing (NLP)
20. What makes semantic analysis more challenging than syntactic processing in NLP?
21. How does discourse processing contribute to understanding context in human
language?
Learning in AI
22. How does inductive learning differ conceptually from explanation-based learning?
23. Why are decision trees a popular choice for learning, and what are their key limitations?
Expert Systems
24. How do expert systems emulate human decision-making in specific domains?
25. Why is knowledge acquisition considered a bottleneck in expert system development?
Set 3 [HARDER QUESTIONS]
Advanced Search Algorithms and Optimization
1. Question: In the context of adversarial search, explain how the Expectimax algorithm
differs from the Minimax algorithm. Under what circumstances is Expectimax more
appropriate, and how does it handle uncertainty in game outcomes?
2. Question: Discuss the limitations of traditional gradient descent methods in training deep
neural networks. How do optimization techniques like Adam and RMSprop address
these limitations?
Probabilistic Graphical Models
3. Question: Define a Hidden Markov Model (HMM) and describe its
components[important]. How would you apply the Forward-Backward algorithm to
compute the posterior probabilities of hidden states given a sequence of
observations[From the subject: NEURAL NETWORK]?
4. Question: Explain the concept of Markov Blanket in Bayesian Networks. How does it
facilitate efficient probabilistic inference, and why is it crucial for understanding the
independence properties within the network?
Reinforcement Learning
5. Question: Compare and contrast Model-Free and Model-Based approaches in
Reinforcement Learning. Provide examples of algorithms for each approach and discuss
their respective advantages and disadvantages in various environments.
6. Question: Describe the Exploration-Exploitation trade-off in Reinforcement Learning.
How do strategies like ε-greedy and Upper Confidence Bound (UCB) address this trade-
off?
Natural Language Processing (NLP)
7. Question: In the realm of NLP, what are Transformer architectures, and how have they
revolutionized tasks such as machine translation and text summarization? Discuss the
significance of self-attention mechanisms within Transformers.
8. Question: Explain the concept of Word Embeddings in NLP. How do models like
Word2Vec and GloVe capture semantic relationships between words, and what are the
limitations of these embeddings?[OUT OF SYLLABUS]
Ethics and AI [Just read once- NOT IN SYLLUBAS]
9. Question: Analyze the ethical implications of deploying AI systems in decision-making
processes that significantly impact human lives, such as in criminal justice or healthcare.
What measures can be implemented to mitigate biases and ensure fairness?
10. Question: Discuss the concept of Explainable AI (XAI). Why is interpretability important
in AI models, especially in critical applications, and what are some methods used to
achieve it?
