(SEM V) THEORY EXAMINATION 2024-25 ARTIFICIAL INTELLIGENCE

Subject Code: BCIT054
Maximum Marks: 70
Time: 3 Hours
Paper ID: 310318

Question Paper Overview

SECTION A (2 × 7 = 14 Marks)

(Short-answer questions — conceptual and definition-based)

a. Define Artificial Intelligence (AI).
b. What is a deterministic environment in AI?
c. Define problem-solving in the context of AI.
d. Define an optimization problem in AI.
e. What is First Order Predicate Logic (FOPL)?
f. Define negotiation in the context of software agents.
g. Explain the concept of machine translation.

SECTION B (Attempt any three × 7 = 21 Marks)

(Descriptive questions — applied and theoretical concepts)

a. Discuss the role of sensors and actuators in intelligent agent systems with examples.
b. Explain the difference between uninformed and informed search strategies with examples.
c. Explain the categories and objects in ontological engineering with examples.
d. Describe the negotiation process in multi-agent systems with examples.
e. Explain the architecture and functioning of speech recognition systems.

SECTION C (Attempt one part from each question × 7 = 35 Marks)

Q3

(a) Explain the role of AI in tackling global challenges like climate change and resource management.
OR
(b) Explain the role of AI in enhancing decision-making processes in business applications.

Q4

(a) Discuss the structure and components of a Constraint Satisfaction Problem (CSP).
OR
(b) Discuss the alpha–beta pruning technique and how it improves game search efficiency.

Q5

(a) Compare semantic networks and frames as knowledge representation methods.
Explain how FOPL is used for representing knowledge in AI systems.
OR
(b) Explain how FOPL is used for representing knowledge in AI systems.

Q6

(a) Explain how argumentation is used for resolving conflicts in multi-agent systems.
OR
(b) Discuss the use of trust metrics in multi-agent decision-making.

Q7

(a) Discuss the challenges of multilingual support in machine translation systems.
OR
(b) Explain how tokenization impacts natural language processing (NLP) tasks.

Key Topics for Revision

1. Artificial Intelligence – Overview

Definition: The simulation of human intelligence processes by machines, especially computer systems.

Major AI Fields: Machine Learning, NLP, Robotics, Computer Vision, Expert Systems, and Reasoning.

Applications: Self-driving cars, chatbots, predictive analytics, fraud detection, smart assistants.

2. Intelligent Agents

Definition: An agent perceives its environment via sensors and acts upon it through actuators.

Example:

Vacuum cleaner robot → sensors (dirt detectors), actuators (wheels).

Self-driving car → cameras, radar, GPS as sensors; engine & steering as actuators.

3. Search Strategies

Uninformed Search: No domain knowledge used.

Examples: BFS, DFS, Uniform Cost Search.

Informed Search: Uses heuristics for guidance.

Examples: Best-First, A*, Greedy Search.

A Algorithm:*

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

where g(n)g(n)g(n) = cost to reach node, h(n)h(n)h(n) = heuristic estimate to goal.

4. Constraint Satisfaction Problems (CSPs)

Components:

Variables → {X₁, X₂, …, Xₙ}

Domains → possible values for each variable

Constraints → relations restricting variable combinations

Examples: N-Queens, Sudoku, Map Coloring.

Solution: A complete assignment satisfying all constraints.

5. Alpha–Beta Pruning

Purpose: Optimize minimax algorithm by pruning irrelevant branches.

Principle:

α (alpha) → Best (max) value found so far.

β (beta) → Best (min) value found so far.

Stop exploring when α ≥ β.

Result: Same decision as Minimax but with fewer node expansions.

6. Knowledge Representation (KR)

Methods:

Semantic Networks: Graph structure (nodes = concepts, edges = relationships).

Frames: Data structures storing object attributes and default values.

FOPL (First Order Predicate Logic):

Represents objects, relations, and quantifiers.

Syntax: ∀x (Human(x) → Mortal(x))

Expresses rules and facts logically.

7. Multi-Agent Systems

Negotiation:

Agents cooperate or compete to reach agreements.

Techniques: Bidding, contract net, argumentation.

Argumentation: Used for conflict resolution based on reasoning and evidence.

Trust Metrics: Quantify agent reliability for cooperation and decision-making.

8. Ontological Engineering

Ontology: Formal representation of knowledge with categories, relations, and rules.

Components:

Classes: Concepts (e.g., “Vehicle”).

Instances: Objects (e.g., “Car1”).

Relations: Links between concepts (e.g., “hasPart”).

Used in semantic web, intelligent agents, and NLP systems.

9. Speech Recognition Systems

Architecture:

Acoustic Model → maps audio to phonemes.

Language Model → predicts word sequences.

Decoder → selects the most probable text output.

Applications: Siri, Alexa, voice search, dictation.

10. Machine Translation (MT)

Definition: Automatic translation of text/speech between languages.

Approaches:

Rule-Based MT (RBMT)

Statistical MT (SMT)

Neural MT (NMT)

Challenges:

Multilingual support, context preservation, idiomatic expressions.

Example: Google Translate, DeepL.

11. Tokenization in NLP

Definition: Splitting text into smaller units (tokens → words, subwords, sentences).

Importance:

Basis for parsing, sentiment analysis, and machine translation.

Affects performance in transformer models (e.g., BERT, GPT).

12. AI for Global Challenges

Climate Change: Predictive analytics for weather, carbon tracking, disaster management.

Resource Management: AI-based water and energy optimization, waste sorting automation.

13. AI in Business Decision-Making

AI helps in data-driven insights, forecasting, and process automation.

Examples: Recommendation systems, risk management, predictive maintenance.

14. Argumentation & Trust in Multi-Agent Systems

15. Important Definitions for Section A