AI Question Paper

Q1) Solve any Six of the following: [6×2=12]

a) What is knowledge and data?

Ans:

 Data: Raw facts or observations without context (e.g., numbers, text, or symbols). Example:
"23, apple, blue."

 Knowledge: Processed and structured data that enables understanding or decision-making.

Example: "The apple weighs 23 grams and is blue."

b) Explain AI benefits.
Ans:

1. Automation: AI can automate repetitive tasks, increasing efficiency.

2. Decision Support: AI analyzes large datasets to make informed decisions.

3. Improved Accuracy: AI reduces human error in tasks like diagnosis and predictions.

4. Innovation: AI enables new technologies, like self-driving cars and virtual assistants.

c) Define Backward chaining.

Ans:
Backward chaining is an inference method in AI where the system starts with the goal and works
backward to determine the required conditions to achieve it.
Example: Diagnosing a disease based on symptoms by hypothesizing possible diseases and
confirming with evidence.

d) List AI applications.
Ans:

1. Virtual Assistants (e.g., Siri, Alexa).

2. Autonomous Vehicles.

3. Fraud Detection in Banking.

4. Medical Diagnosis Systems.

5. Recommendation Systems (e.g., Netflix, Amazon).

e) Describe any two characteristics of a problem.

Ans:
 1. Well-defined Initial State: A clear starting point, e.g., the starting position in a chess game.

2. Goal State: A specific outcome or solution, e.g., solving a puzzle.

f) What is Classical Planning in AI?

Ans:
Classical planning involves generating a sequence of actions from an initial state to a goal state,
assuming a deterministic and fully observable environment. Example: Planning robot movement in a
factory.

g) What is an Intelligent Agent?

Ans:
An intelligent agent is an autonomous system that perceives its environment through sensors and
acts on it using actuators to achieve specific goals.
Example: A robotic vacuum cleaner.

Q2) Attempt any Four of the following: [4×3=12]

a) Explain Water Jug Problem in detail.

Ans:
The Water Jug Problem is a classic problem-solving scenario where two jugs of different capacities
are used to measure a specific amount of water.

Example:

 Jug X: 4 liters, Jug Y: 3 liters.

 Goal: Measure exactly 2 liters.

Solution Steps:

1. Fill Jug Y completely.

2. Pour water from Jug Y into Jug X until Jug X is full.

3. Empty Jug X and pour remaining 1 liter from Jug Y into Jug X.

4. Fill Jug Y again and pour into Jug X until it contains 2 liters.

b) Describe the A algorithm.*

Ans:
The A Algorithm* is a pathfinding and graph traversal algorithm that uses:

 g(n): Cost from the start node to the current node.

 h(n): Estimated cost from the current node to the goal (heuristic).

 f(n): Total estimated cost: f(n)=g(n)+h(n)f(n) = g(n) + h(n).

Steps:
 1. Start with the initial node.

2. Expand the node with the lowest f(n)f(n).

3. Update costs for adjacent nodes and repeat until the goal is reached.

c) Discuss in brief the various issues in Knowledge Representation.

Ans:

1. Expressiveness: The representation must be able to capture complex real-world knowledge.

2. Consistency: The representation should avoid contradictions in knowledge.

3. Scalability: Must handle large volumes of knowledge efficiently.

4. Inference: Should enable efficient reasoning and deduction.

d) Explain the MINIMAX procedure.

Ans:
MINIMAX is a decision-making algorithm used in two-player games.

Steps:

1. Assume two players: Maximizer and Minimizer.

2. Generate a game tree of all possible moves.

3. Evaluate leaf nodes with a heuristic function.

4. Backpropagate values to parent nodes, alternating between maximizing and minimizing

scores.

Example:
Used in chess to decide the best move.

e) Convert to FOPL and prove:

1. John likes all kinds of food.

2. Apple and vegetables are food.

3. Anil eats peanuts and is still alive.

Ans:
FOPL Statements:

1. ∀x(Food(x)→Likes(John,x))

2. Food(Apple)∧Food(Vegetables)Food(Apple)

3. Eats(Anil,Peanuts)∧Alive(Anil)Eats(Anil, Peanuts)
Proof:
Given the above statements, we can infer:

 Likes(John,Apple)Likes(John, Apple) and Likes(John,Vegetables)Likes(John, Vegetables).

 No contradiction exists for Anil eating peanuts and being alive.

Q3) Attempt any Three of the following: [3×4=12]

a) Explain Hill Climbing Algorithm in detail.

Ans:
Hill Climbing is a local search algorithm that iteratively moves towards the goal by choosing the
neighbor with the highest value.

Steps:

1. Start with an initial state.

2. Evaluate neighboring states.

3. Move to the neighbor with the best heuristic value.

4. Repeat until the goal or a local maximum is reached.

Limitation:
May get stuck in local maxima, plateaus, or ridges.

b) Explain Monte Carlo Tree Search technique.

Ans:
Monte Carlo Tree Search (MCTS) is a search algorithm used in decision-making processes like games.

Steps:

1. Selection: Traverse the tree using a selection policy like UCB1.

2. Expansion: Add a new child node.

3. Simulation: Perform random playouts from the new node.

4. Backpropagation: Update the node's statistics based on the result.

c) Define Constraint Satisfaction Problem (CSP). Solve SEND+MORE=MONEY.

Ans:
CSP Definition:
A constraint satisfaction problem (CSP) is a mathematical problem that involves assigning values to
variables within a set of constraints

SEND + MORE = MONEY Solution:

Assign unique digits to letters so that the equation holds true.

Solution:
S=9, E=5, N=6, D=7, M=1, O=0, R=8, Y=2.

Verification:

9567 + 1085 = 10652

Q4) Attempt any Three of the following: [3×4=12]

a) Explain Means-Ends Analysis Algorithm in detail.

Ans:
Means-Ends Analysis (MEA) is a problem-solving technique used in AI to reduce the difference
between the current state and the goal state by identifying and applying appropriate actions.

Steps of the Algorithm:

1. Initial State Comparison:

o Compare the current state with the goal state to identify the differences.
 2. Select an Operator:

o Choose an operator that can reduce one or more differences.

3. Subgoals Generation:

o If the operator cannot be applied directly, define subgoals to satisfy the

preconditions of the operator.

4. Apply the Operator:

o Once preconditions are satisfied, apply the operator to reduce the difference.

5. Iterate:

o Repeat the process until the current state matches the goal state.

Example:
Goal: Move from Pune to Mumbai.

 Current State: In Pune.

 Goal State: In Mumbai.

 Operators: Travel by bus, train, or plane.

1. Compare states: Pune ≠ Mumbai.

2. Choose the operator: Travel by train.

3. Apply the operator: Book train tickets and board the train.

b) Apply Alpha-Beta Pruning Algorithm for the following search.

Ans:
Alpha-Beta Pruning is a technique to optimize the MINIMAX algorithm by skipping branches that
won't affect the final decision.

Example Search Tree:

MAX

/ \

MIN MIN

/|\ /|\

3 5 6 7 4 2

Steps:

1. Start at the root node (MAX).

2. Explore the left subtree (MIN) first.

o Compare 3, 5, and 6. The MIN value is 3.

3. Prune unnecessary nodes in the right subtree (MIN).

 o Since 3 (left subtree) is better than any potential value from the right subtree, skip
further evaluation.

4. Return 3 as the optimal value.

c) Describe any two approaches of representing knowledge.

Ans:

1. Semantic Networks:

o Represent knowledge as a graph with nodes (concepts or objects) and edges

(relationships).

o Example: "A dog is an animal" is represented as a node "Dog" connected to "Animal"

via an "is-a" link.

2. Production Rules:

o Represent knowledge as condition-action pairs.

o Example:

o IF temperature > 30°C

o THEN turn on the fan.

d) Explain Mental and Modal Object Logics.

Ans:

1. Mental Logic:

o Refers to the formal representation of reasoning processes in human cognition.

o Focuses on mental states like beliefs, desires, and intentions.

Example:
Belief representation: B(A,P) means "Agent A believes proposition P."

2. Modal Logic:

o A formal system for representing modalities like possibility and necessity.

o Syntax uses operators:

 □P: Necessarily PP.

 ◊P: Possibly PP.

Example:

 □(P→Q) "If PP, then QQ is necessarily true."

 Used in AI for reasoning about uncertainty and hypothetical scenarios.

Q5) Attempt the following:

a) Attempt any TWO of the following. [2×5=10]

i) What is Hierarchical Planning? Explain in detail.

Ans:
Hierarchical Planning is a planning technique that involves decomposing high-level goals into
smaller, more manageable subgoals or tasks. This approach uses a hierarchy of goals and actions,
making the planning process more efficient, particularly in complex systems.

Steps Involved:

1. Goal Decomposition: A high-level goal is decomposed into subgoals or tasks.

2. Subtask Assignment: Each subgoal is further broken down into more specific actions until
the problem becomes solvable at a low level.

3. Action Selection: Choose appropriate actions to achieve the subgoals, often considering
constraints and resources.

4. Execution: Execute the low-level actions, and replan as needed.

Example:

 High-Level Goal: Get to the office.

 Subgoals:

o Travel to the bus stop.

o Take the bus to the office.

 Decomposing Further:

o For "Travel to bus stop," tasks may include "Leave house," "Walk to bus stop."

Hierarchical planning is useful in AI applications such as robotics, where complex tasks can be broken
down into smaller, simpler tasks.

ii) Explain First-Order Logic with example.

Ans:
First-Order Logic (FOL) is a formal system used for knowledge representation and reasoning in AI.
FOL extends propositional logic by incorporating variables, predicates, and quantifiers, making it
more expressive.

Components:

1. Predicates: Describe properties or relationships about objects (e.g., P(x)P(x) means "x is a
person").
 2. Constants: Specific objects or entities (e.g., JohnJohn is a constant).

3. Variables: Symbols that can take on values (e.g., xx, yy).

4. Quantifiers:

o Universal quantifier (∀\forall): "For all" (e.g., ∀xP(x)\forall x P(x) means "All x are
persons").

o Existential quantifier (∃\exists): "There exists" (e.g., ∃xP(x)\exists x P(x) means

"There exists some x who is a person").

Example:

 ∀x(Human(x)→Mortal(x))\forall x (Human(x) \rightarrow Mortal(x)) means "All humans are

mortal."

 ∃x(Human(x)∧¬Mortal(x))\exists x (Human(x) \land \neg Mortal(x)) means "There exists a

human who is not mortal" (although logically false in our world).

FOL is used extensively in AI for reasoning, theorem proving, and knowledge representation.

iii) State components using which Problem can be formatted?

Ans:
A problem in AI can typically be formulated using the following components:

1. Initial State: The starting point or configuration of the environment at the beginning of the
problem.
Example: In a maze, the initial state would be the starting position of the agent.

2. Actions: A set of operators or moves that transform the current state into a new state.
Example: Moving left, right, up, or down in a maze.

3. Transition Model: Describes the results of applying actions to states (i.e., how actions lead
from one state to another).
Example: The result of moving up in a maze might change the agent’s position.

4. Goal State: The desired state that represents the solution to the problem.
Example: Reaching the exit in a maze.

5. Path Cost: A measure of the "cost" of a particular path from the initial state to the goal state.
Example: The time or distance it takes to move through the maze.

b) Attempt any ONE of the following. [1×2=2]

i) Define Search Strategy terminology.

Ans:
A search strategy in AI refers to the method used to explore the search space (the set of all possible
states or configurations) in order to find a solution to the problem. The main components of a search
strategy include:
 1. Node: A state or configuration in the search space.

2. State Space: The collection of all possible states the problem can take.

3. Search Tree: A tree-like structure representing the states and transitions between them.

4. Search Frontier: The set of nodes that have been generated but not yet expanded.

5. Goal Test: A function that checks if a state is the goal state.

6. Search Algorithm: The procedure or method (e.g., Depth-First Search, Breadth-First Search,
A*) used to explore the search space.

ii) Explain Language Model in AI.

Ans:
A Language Model in AI is a probabilistic model used to predict the likelihood of a sequence of words
or phrases in natural language processing (NLP). It is used in applications such as speech recognition,
text generation, and machine translation.

Types of Language Models:

1. N-Gram Model: Predicts the next word based on the previous N-1 words (e.g., bigrams,
trigrams).

2. Neural Network-based Models: Uses deep learning models (like RNNs, LSTMs) to learn
complex patterns in text data and generate more accurate predictions.

Language models are trained using large corpora of text, and the model is evaluated based on how
well it predicts the likelihood of unseen sequences.

Q6) Write a short note. Any Two. [2×5=10]

a) Automated Planning.
Ans:
Automated Planning is the process in AI where a system autonomously creates a sequence of
actions (plan) to achieve a specific goal. It involves the formulation of a plan from an initial state to a
goal state.

Key Components:

1. Initial State: Where the planning process begins.

2. Actions: The set of operators that can be applied.

3. Goal: The desired final state.

4. Plan: The sequence of actions that will lead to the goal.

Types of Planning:

 Classical Planning: Assumes fully observable and deterministic environments.

  Conditional Planning: Accounts for uncertain outcomes and different contingencies.

Automated planning is widely used in robotics, logistics, and autonomous systems.

b) AO Algorithm.*
Ans:
AO (And-Or A)** is a search algorithm used for solving AND-OR graphs, which involve both "AND"
and "OR" nodes, making it suitable for problems where tasks have both dependencies (AND) and
alternatives (OR).

Steps Involved:

1. Nodes: Represent tasks or subgoals.

o AND Nodes: Multiple conditions must be satisfied to reach a goal.

o OR Nodes: Only one condition needs to be satisfied.

2. Search: AO* recursively expands the graph, evaluating costs based on both the AND and OR
relationships.

3. Solution: The algorithm seeks the least-cost path by considering the structure of the graph
and pruning unnecessary branches.

AO* is used in complex decision-making problems where tasks are interdependent.

c) Information Extraction.
Ans:
Information Extraction (IE) refers to the process of automatically extracting structured information
from unstructured data, typically text. It is a subfield of natural language processing (NLP) that
focuses on identifying entities, relationships, and events from textual data.

Common IE tasks:

1. Named Entity Recognition (NER): Identifying entities such as names, dates, locations.

2. Relation Extraction: Identifying relationships between entities (e.g., person X works for
company Y).

3. Event Extraction: Detecting events or actions described in the text.

Applications:

 News aggregation and summarization.

 Bioinformatics (e.g., extracting genes, proteins).

 Customer support automation.

IE systems use techniques like rule-based methods, machine learning, and deep learning to process
and extract meaningful information.