BE Computer Engineering (Pokhara University) Artificial Intelligence (PU, CMP 346) Question Paper 2078
This is the official BE Computer Engineering (Pokhara University) Artificial Intelligence (PU, CMP 346) question paper for 2078, as set in the regular annual examination. It carries 100 full marks and a time allowance of 180 minutes, across 12 questions. On Kekkei you can attempt this Artificial Intelligence (PU, CMP 346) past paper online with a timer, get instant AI feedback and step-by-step solutions, and track the topics where you lose marks — completely free. Whether you are revising for your BE Computer Engineering (Pokhara University) Artificial Intelligence (PU, CMP 346) exam or solving previous years' question papers, this 2078 paper is a great way to practise under real exam conditions.
Section A: Long Answer Questions
Attempt all / any as specified.
(a) Define a rational agent and explain the PEAS (Performance measure, Environment, Actuators, Sensors) framework used to specify a task environment. (6 marks)
(b) An autonomous vacuum-cleaning robot operates in a two-room environment. Give the PEAS description for this agent. (4 marks)
(c) Classify a task environment along the following dimensions and justify your classification for the vacuum-cleaning robot: fully vs. partially observable, deterministic vs. stochastic, episodic vs. sequential, and single vs. multi-agent. (4 marks)
(a) Distinguish between uninformed and informed (heuristic) search strategies with suitable examples. (4 marks)
(b) State the A* search evaluation function f(n) = g(n) + h(n) and define each term. Explain what is meant by an admissible heuristic and a consistent (monotonic) heuristic. (6 marks)
(c) Prove that A* tree search is optimal when the heuristic h(n) is admissible. (6 marks)
(a) Explain the resolution inference rule and outline the steps to convert a propositional sentence into Conjunctive Normal Form (CNF). (6 marks)
(b) Using resolution refutation, prove that the following knowledge base entails Q:
- P ⇒ Q
- (L ∧ M) ⇒ P
- (B ∧ L) ⇒ M
- (A ∧ P) ⇒ L
- (A ∧ B) ⇒ L
- A
- B
(8 marks)
(a) Draw the architecture of a single artificial neuron (perceptron) and describe the role of weights, bias and activation function. (5 marks)
(b) Explain the working of the backpropagation algorithm used to train a multilayer feedforward neural network, clearly describing the forward pass, error computation and weight-update phases. (7 marks)
Section B: Short Answer Questions
Attempt all / any as specified.
Compare Breadth-First Search and Depth-First Search in terms of completeness, optimality, time complexity and space complexity. State one situation where each is preferred.
What is knowledge representation? Briefly explain semantic networks and frames as knowledge-representation schemes, mentioning one advantage and one limitation of each.
Draw and explain the basic architecture of an expert system. Differentiate between forward chaining and backward chaining used by its inference engine.
Represent the following English statements in First-Order Predicate Logic:
(a) Every student who studies passes the exam. (b) Some birds cannot fly. (c) All cats are animals. (d) John loves everyone who loves Mary.
Explain the Hill-Climbing search algorithm. Discuss the problems of local maxima, plateaus and ridges, and state one technique to overcome each.
Differentiate between supervised, unsupervised and reinforcement learning with one example application of each.
What is Natural Language Processing (NLP)? Briefly explain the phases of NLP (lexical, syntactic, semantic, discourse and pragmatic analysis).
Write short notes on any TWO of the following:
(a) Simple reflex agent (b) Model-based reflex agent (c) Goal-based agent (d) Utility-based agent