Python Programming for Data Science

Question Bank
Module -1
Sl Question Mar Bloo
N ks m’s
O Level
1. What is Artificial Intelligence? Explain its evolution. 8 L2
 Definition of AI (1 mark):

 AI refers to the simulation of human intelligence in machines

programmed to think and learn like humans.

 Core components of AI (1 mark):

 Machine learning, natural language processing, robotics, and

expert systems.

 Early beginnings (1 mark):

 Originated in the 1950s with pioneers like Alan Turing and John
McCarthy.
 The Dartmouth Conference (1956) coined the term "Artificial
Intelligence."

 First AI Winter (1974-1980) (1 mark):

 Reduced funding and interest due to limitations of early AI

systems.

 Expert Systems Era (1980s) (1 mark):

 Focus on rule-based systems mimicking human expertise in

specific domains.

 Machine Learning Revolution (1990s-present) (1 mark):

 Shift towards data-driven approaches and statistical learning.

 Development of neural networks and deep learning techniques.

 Modern AI advancements (1 mark):

  Breakthroughs in natural language processing, computer vision,
and robotics.
 AI integration in various sectors: healthcare, finance, automotive,
etc.

 Current trends and future outlook (1 mark):

 Focus on ethical AI, explainable AI, and artificial general

intelligence (AGI).
 Ongoing research in areas like reinforcement learning and
quantum AI.

2. Illustrate the importance of turing test and state how this can be considered a 5 L2
benchmark for any AI algorithms
 Definition and concept (1 mark):

 The Turing Test, proposed by Alan Turing in 1950, is a method to

assess a machine's ability to exhibit intelligent behavior
indistinguishable from a human.

 Importance in AI development (1 mark):

 Serves as a philosophical and practical framework for defining

machine intelligence.
 Encourages the development of AI systems that can understand
and generate human-like responses.

 Benchmark for natural language processing (1 mark):

 Challenges AI to master complex aspects of human

communication, including context, nuance, and cultural
references.
 Drives improvements in chatbots, virtual assistants, and language
models.

 Limitations and criticisms (1 mark):

 Focuses primarily on linguistic performance, potentially

neglecting other aspects of intelligence.
 Some argue it tests for human-like behavior rather than true
understanding or consciousness.

 Ongoing relevance in modern AI (1 mark):

 While not the sole measure of AI capability, it remains a widely

recognized concept in the field.
 Variations like the Winograd Schema Challenge have been
 developed to address some limitations.

3. State the various types of intelligence with suitable examples for each category. 8 L2
the various types of intelligence with suitable examples for each category:

1. Linguistic Intelligence:
o Ability to use language effectively, both verbally and in
writing
o Examples: Writers, poets, journalists, speakers, translators
2. Logical-Mathematical Intelligence:
o Capacity to understand and work with numbers, logical
reasoning, and abstractions
o Examples: Mathematicians, scientists, engineers,
accountants, programmers
3. Spatial Intelligence:
o Skill in visualizing and manipulating objects and spatial
dimensions
o Examples: Architects, artists, navigators, chess players,
surgeons
4. Musical Intelligence:
o Ability to produce and appreciate rhythm, pitch, and
timbre
o Examples: Musicians, composers, conductors, music
critics
5. Bodily-Kinesthetic Intelligence:
o Control of one's body movements and the capacity to
handle objects skillfully
o Examples: Athletes, dancers, actors, craftspeople, surgeons
6. Interpersonal Intelligence:
o Capacity to understand and interact effectively with others
o Examples: Teachers, counselors, politicians, salespeople,
managers
7. Intrapersonal Intelligence:
o Self-awareness and the ability to understand one's own
emotions, motivations, and inner states
o Examples: Psychologists, philosophers, spiritual leaders
8. Naturalistic Intelligence:
o Ability to recognize and categorize plants, animals, and
other elements of nature
o Examples: Biologists, ecologists, botanists,
environmentalists
9. Existential Intelligence:
o Sensitivity and capacity to tackle deep questions about
human existence
o Examples: Philosophers, theologians, cosmologists
10. Emotional Intelligence:
o Ability to perceive, control, and evaluate emotions
 o Examples: Therapists, negotiators, effective leaders

4. Examine the AI literature to discover whether or not the following tasks can 10 L4
currently be solved by computers:
a. Playing a decent game of table tennis (ping-pong).
b. Giving tablets to a bedridden patient.
c. Discovering and proving new mathematical theorems.
e. Writing an intentionally funny story.
f. Giving competent legal advice in a specialized area of law. For the currently
infeasible tasks, try to find out what the difficulties are and estimate when they
will be overcome
Playing a decent game of table tennis (ping-pong):

Currently feasible: Yes Explanation: Robots capable of playing table

tennis have been developed. For example, FORPHEUS by Omron
Corporation can play against humans and even coach them. These
systems use high-speed cameras, motion sensors, and advanced
algorithms to track the ball and opponent movements, calculate
trajectories, and execute precise movements.

b. Giving tablets to a bedridden patient:

Currently feasible: Partially Explanation: While robots can perform

simple tasks in healthcare settings, the complexity of safely interacting
with a bedridden patient presents challenges. Difficulties:

1. Safe physical interaction with fragile patients

2. Accurate patient recognition and status assessment
3. Adaptability to various patient positions and room layouts
Estimated timeline: Fully capable systems may be available by
2028-2030, as advances in soft robotics, computer vision, and
human-robot interaction progress.

c. Discovering and proving new mathematical theorems:

Currently feasible: Partially Explanation: AI systems have made

significant strides in this area. For instance, the Lean theorem prover has
been used to formalize complex mathematical proofs. However,
discovering entirely new, significant theorems remains a challenge.
Difficulties:

1. Creative insight required for novel discoveries

2. Understanding the significance and context of new theorems
3. Explaining proofs in a way that's comprehensible to humans
Estimated timeline: Significant breakthroughs in AI-driven
mathematical discovery may occur by 2030-2035, as systems
become more adept at abstract reasoning and creativity.

e. Writing an intentionally funny story:

 Currently feasible: Partially Explanation: AI language models can
generate humorous content, but creating consistently funny and
intentional humor remains challenging. Difficulties:

1. Understanding context, cultural references, and subtleties of

human humor
2. Generating original, clever jokes rather than recycling known
patterns
3. Maintaining coherence and narrative structure while being funny
Estimated timeline: More sophisticated humor-generating AI may
emerge by 2026-2028, but truly matching human-level humor
creation could take until 2030 or beyond.

f. Giving competent legal advice in a specialized area of law:

Currently feasible: Partially Explanation: AI systems can assist with legal

research and provide information on legal precedents, but giving nuanced,
context-aware legal advice is still primarily a human domain. Difficulties:

1. Interpreting complex, sometimes ambiguous legal language

2. Understanding the full context of a case, including non-textual
factors
3. Applying ethical considerations and professional judgment
4. Keeping up with rapidly changing laws and precedents Estimated
timeline: AI systems capable of providing competent legal advice
in specialized areas may emerge by 2028-2030, but they will
likely work alongside human lawyers rather than replacing them
entirely.

For all these tasks, it's important to note that while AI can make
significant progress, human oversight and collaboration will likely remain
crucial for the foreseeable future, especially in sensitive areas like
healthcare and law.

5 Explain in detail the architecture for designing intelligent agent. 8 L2

In general,
the architecture makes the percepts from the sensors available to the program, runs the
program,
and feeds the program's action choices to the effectors as they are generated. The
relationship
among agents, architectures, and programs can be summed up as follows:
agent = architecture + program
Before we design an agent program, we must have a pretty good idea of the possible
percepts and actions, what goals or performance measure the agent is supposed to
achieve, and
what sort of environment it will operate in.5 These come in a wide variety
6 While driving, which is the best policy? 10 L4
a. Always put your directional blinker on before turning,
b. Never use your blinker,
c. Look in your mirrors and use your blinker only if you observe a car that can
observe you?
What kind of reasoning did you need to do to arrive at this policy (logical, goal-
based, or utilitybased)?
What kind of agent design is necessary to carry out the policy (reflex, goal-
based, or utility-based)?
a. Always put your directional blinker on before turning b. Never use your
blinker c. Look in your mirrors and use your blinker only if you observe a
car that can observe you

The best policy is:

a. Always put your directional blinker on before turning

Reasoning:

This is utility-based reasoning. Here's why:

1. Safety: Using your blinker always maximizes safety for all road
users.
2. Predictability: It makes your actions predictable to other drivers,
pedestrians, and cyclists.
 3. Legal compliance: It's typically required by law in most
jurisdictions.
4. Habit formation: Consistently using your blinker builds a good
habit.
5. Consideration for unseen road users: There may be road users you
haven't noticed.

Option b is clearly unsafe and illegal. Option c, while seemingly logical,

fails to account for road users you might not see and doesn't promote
consistent, habitual safe driving.

The reasoning is utility-based because it considers the outcomes and

benefits (utility) of each action for all stakeholders, not just the driver. It
goes beyond simple logical rules or specific goals, evaluating the overall
best outcome in various scenarios.

Agent design necessary to carry out the policy:

A reflex agent design is sufficient to carry out this policy. Here's why:

1. Simple rule: The policy "always use your blinker before turning"
is a straightforward rule that can be implemented as a reflex
action.
2. No complex decision-making: The agent doesn't need to evaluate
goals or utilities in real-time; it simply needs to activate the
blinker before any turn.
3. Direct mapping from perception to action: When the agent
perceives that a turn is imminent, it can directly trigger the action
of using the blinker.
4. No internal state required: The agent doesn't need to maintain any
internal state or model of the world to implement this policy.

While a more complex agent (goal-based or utility-based) could certainly

implement this policy, it's not necessary. A simple reflex agent that
activates the blinker in response to the perception of an upcoming turn is
sufficient to consistently carry out this safe driving practice.
7. Illustrate the various categories of agent programs available and explain each 8 L2
category with suitable real time application.

Categories of Agent Programs

Agent programs are the core of intelligent agents, determining how they
process percepts and decide on actions. Here are the main categories of
agent programs:

1. Simple Reflex Agents

Explanation: These agents select actions based on the current percept,
ignoring any history. They use condition-action rules to map percepts
directly to actions.

Real-time Application: Traffic light control system

 Percept: Camera detects traffic flow

 Action: Change light based on predefined rules (e.g., if long
queue, change to green)

2. Model-Based Reflex Agents

Explanation: These agents maintain an internal state to track aspects of
the world not visible in the current percept. They use this state along with
the current percept to choose actions.

Real-time Application: Autonomous vacuum cleaner

 Internal State: Map of cleaned areas

 Percept: Current location, obstacles
 Action: Move to uncleaned areas, avoid obstacles

3. Goal-Based Agents
Explanation: These agents consider future consequences of actions. They
use goal information to decide which situations are desirable.

Real-time Application: GPS navigation system

 Goal: Reach destination

 Action: Plan route, provide turn-by-turn directions
 Adapts to traffic conditions, road closures to achieve goal

4. Utility-Based Agents
Explanation: These agents have a more general performance measure
(utility function) to evaluate different possible scenarios, allowing for
more complex decision-making.

Real-time Application: Smart energy management system

 Utility Function: Balance comfort and energy efficiency

 Percepts: Room temperature, electricity prices, user preferences
 Action: Adjust heating/cooling to maximize utility

5. Learning Agents
Explanation: These agents can improve their performance over time
through experience. They have a learning component that modifies other
 components to improve overall performance.

Real-time Application: Recommender system (e.g., Netflix, Amazon)

 Learning Component: Analyzes user behavior and feedback

 Performance Element: Recommends content or products
 Improves recommendations over time based on user interactions

6. Hybrid Agents
Explanation: These agents combine two or more of the above
architectures to leverage the strengths of each.

Real-time Application: Autonomous vehicle

 Reflex Component: Emergency braking for sudden obstacles

 Model-Based Component: Tracking other vehicles' positions
 Goal-Based Component: Navigation to destination
 Utility-Based Component: Optimizing for fuel efficiency, time,
and passenger comfort
 Learning Component: Improving driving behavior over time

Each category of agent program has its strengths and is suited to different
types of environments and problems. The choice of agent program
depends on the complexity of the environment, the nature of the task, and
the resources available for implementation.
8 Explain the various properties of Environments available. 8 L2

Properties of Environments in AI
Understanding the properties of environments is crucial for designing
effective intelligent agents. Here are the key properties of environments:

1. Fully Observable vs. Partially Observable

 Fully Observable: The agent can obtain complete, accurate, up-
to-date information about the environment's state at each point in
time.
o Example: Chess game (the entire board is visible)
 Partially Observable: The agent's sensors cannot detect all
aspects of the environment at each time step.
o Example: Poker game (opponent's cards are hidden)

2. Deterministic vs. Stochastic

 Deterministic: The next state of the environment is completely
determined by the current state and the agent's action.
 o Example: Puzzle-solving environment
 Stochastic: There's an element of randomness or uncertainty in
how the environment changes in response to an action.
o Example: Weather forecasting system

3. Episodic vs. Sequential

 Episodic: The agent's experience is divided into atomic episodes.
Actions in one episode don't affect other episodes.
o Example: Image classification task
 Sequential: Current decisions affect future situations.
o Example: Chess game (each move affects future game
states)

4. Static vs. Dynamic

 Static: The environment doesn't change while the agent is
deliberating.
o Example: Sudoku puzzle
 Dynamic: The environment can change while the agent is
thinking.
o Example: Real-time strategy game

5. Discrete vs. Continuous

 Discrete: There are a finite number of distinct states and actions.
o Example: Chess (finite number of board positions)
 Continuous: States or actions (or both) form a continuous range
of values.
o Example: Robot arm control (continuous range of angles)

6. Single-agent vs. Multi-agent

 Single-agent: Only one agent operates in the environment.
o Example: Puzzle-solving robot
 Multi-agent: Multiple agents operate in the environment,
potentially interacting or competing.
o Example: Autonomous vehicles in traffic

7. Known vs. Unknown

 Known: The agent knows the rules and outcomes of the
environment.
o Example: Chess game with known rules
 Unknown: The agent must learn how the environment works.
o Example: Robot exploring a new planet
8. Accessible vs. Inaccessible
 Accessible: The agent can obtain information about all states of
the environment directly.
o Example: Database query system
 Inaccessible: The agent needs to maintain an internal state to keep
track of the environment.
o Example: Autonomous car navigating a city

9. Factored vs. Structured

 Factored: The state of the environment can be neatly divided into
independent attributes or variables.
o Example: Weather prediction (temperature, humidity,
pressure as separate factors)
 Structured: The environment has complex relationships between
its components.
o Example: Social network analysis

10. Friendly vs. Adversarial

 Friendly: The environment is not actively trying to hinder the
agent's goals.
o Example: Personal assistant AI
 Adversarial: The environment (or other agents in it) actively
opposes the agent's goals.
o Example: Competitive game-playing AI

Understanding these properties helps in designing appropriate agent

architectures and algorithms. Real-world environments often combine
multiple properties, and the challenge lies in creating agents that can
effectively operate under these complex conditions.
 Module-2

Sl NO Question Marks Bloom’s Level

1. Explain the various search strategies and the evaluation 8 L2
strategies to handle them.
Search strategies are fundamental to problem-solving in AI.
They are used to find solutions in a search space, which
represents all possible states of the problem. Here's an
overview of various search strategies and their evaluation
methods:
There are three main categories of search strategies:

1. Uninformed (Blind) Search Strategies

2. Informed (Heuristic) Search Strategies
3. Local Search Algorithms

I. Uninformed (Blind) Search

Strategies
These strategies don't use domain-specific knowledge
beyond the problem definition.

BFS,DFS,Depth Limited Search, Iterative searching

mechanisms fall under this category.

II. Informed (Heuristic) Search

Strategies
These strategies use problem-specific knowledge to
guide the search.

Best first search, A* search, Iterative deepening

search falls under this category

III. Local Search Algorithms

These algorithms maintain a single current state and
move to neighboring states.

Techniques like Hill Climbing, Simulated Annealing

and Genetic algorithms fall under this category.

Evaluation Strategies
1. Completeness: Does the algorithm guarantee to
find a solution if one exists?
 2. Optimality: Does the algorithm guarantee to
find the best solution?
3. Time Complexity: How does the execution
time grow with problem size?
4. Space Complexity: How does the memory
usage grow with problem size?
5. Branching Factor: Average number of
successors per state.
6. Depth of Solution: Length of the shortest path
to a solution.
7. Heuristic Accuracy: For informed search, how
well does the heuristic estimate the actual cost?

When evaluating search strategies, it's crucial to

consider the specific problem characteristics and
constraints, such as available memory, time limitations,
and the nature of the search space.

2. Consider the following tree: 10 L3

a) Perform both DFS and BFS traversals on this tree, starting

from the root node.
b) At which level (depth) will DFS find node 7 compared to
BFS?
c) How many nodes will each algorithm visit before finding
node 8?
From the given tree structure
a) Perform DFS and BFS Traversals:

- Depth-First Search (DFS):

DFS explores as far as possible along each branch before
backtracking. The typical traversal order is:
1 → 2 → 4 → 8 → backtrack → 5 → backtrack → 3 → 6 → 7

DFS traversal result:1, 2, 4, 8, 5, 3, 6, 7

 - Breadth-First Search (BFS):
BFS explores all nodes at the present level before moving on
to the nodes at the next level.
1→2→3→4→5→6→7→8

BFS traversal result: 1, 2, 3, 4, 5, 6, 7, 8

b) At which level (depth) will DFS find node 7 compared to

BFS?
DFS will visit nodes in a depth-first manner, so it will reach
node 7 after backtracking from nodes 4, 8, and 5, then
visiting nodes 3 and 6. DFS finds node 7 at depth 3.

- BFS, on the other hand, processes level by level. It will find

node 7 while exploring nodes in level 2, so it finds node 7 at
depth 2.

c) How many nodes will each algorithm visit before finding

node 8?

- DFS:
DFS will visit nodes in the following order until it finds
node 8: 1, 2, 4. It will then visit node8. So DFS will visit 3
nodes before finding node 8.

BFS:
BFS will visit all nodes in the current level before
proceeding. The nodes visited before reaching node 8 will
be: 1, 2, 3, 4, 5, 6, 7. So BFS will visit 7 nodes before finding
node 8.

3. Explain the difference between a problem-solving agent and 8 L2

a simple reflex agent.
A problem-solving agent and a simple reflex agent
are both types of agents used in artificial intelligence
(AI), but they differ in how they approach decision-
making and interaction with their environment.

1. Simple Reflex Agent:

A simple reflex agent operates based on condition-

action rules, meaning it reacts to the current state of the
environment (percept) by following predefined rules
without considering the future or past states. It acts
purely on the current situation without any internal
memory or future planning.

 How it works: The agent observes the

environment (perceives) and applies a set of
rules or conditions to decide on an action. It
 doesn’t think about the consequences of its
actions or any long-term goals.
 Key Characteristics:
o No memory: The agent doesn't have any
history of past actions or states.
o No planning: It doesn't predict the outcome
of actions or reason about future states.
o Reactive: The agent reacts instantly to
current conditions.
o Simplicity: It’s suitable for simple
environments where conditions can be
handled directly by predefined rules.

 Example: A thermostat is a simple reflex agent.

If the temperature (percept) falls below a certain
threshold, it turns on the heating. It doesn't
consider long-term energy efficiency or future
temperatures—it only reacts to the current
temperature.

2. Problem-Solving Agent:

A problem-solving agent, on the other hand, operates

by searching for a sequence of actions that leads to a
goal. It uses knowledge about the current state of the
world and considers possible future states and actions.
This agent often uses algorithms such as search
strategies (e.g., breadth-first search, depth-first search)
to explore potential solutions.

 How it works: The agent starts with a clear goal

and a model of the environment. It explores
different actions and states, evaluates them, and
selects a sequence of actions that lead to the
desired goal.
 Key Characteristics:
o Planning: It predicts the outcomes of
different actions and evaluates future states
to achieve a goal.
o Search: It uses search techniques to find a
solution in a state space.
o Goal-oriented: It has a specific objective to
reach and works towards it.
o More complex: This agent is designed for
more complex environments where goals
and multiple actions need to be considered.

 Example: A navigation system in a car is a

problem-solving agent. Given a destination
 (goal), it plans the best route by searching
through possible routes (state space) and
considering factors like distance or traffic. It
doesn't just react but plans ahead to achieve the
goal efficiently.

Key Differences:
Feature Simple Reflex Agent Problem-Solving Agent

Reactive (Condition- Goal-Oriented and

Approach
Action rules) Planning

No memory or state Can store and evaluate

Memory
history states

No planning; acts
Plans a sequence of
Planning based on current
actions based on goals
percept

Simple; works well in More complex; works

Complexity structured well in dynamic,
environments uncertain environments

Thermostat, light Navigation system,

Example
switch chess-playing agent

4. Compare and contrast depth-first search and breadth-first 5 L2

search strategies.
Depth-First Search (DFS) and Breadth-First Search
(BFS) are two fundamental search strategies used to
explore or traverse graphs and trees. While both search
methods are designed to systematically visit nodes, they
do so in very different ways. Here's a detailed
comparison and contrast:

1. Approach:

 DFS (Depth-First Search):

o DFS explores as far as possible along a
branch before backtracking.
o It starts at the root (or an arbitrary starting
node in graphs), moves to one of its
children, and continues along the depth of
the tree/graph until it reaches a node with
no unvisited neighbors. Then, it backtracks
to explore other unvisited paths.

 BFS (Breadth-First Search):

o BFS explores all nodes at the present depth
level before moving on to nodes at the next
level.
 o It starts at the root and visits all its
neighbors (children) first before visiting the
children of those neighbors.

2. Traversal Strategy:

 DFS:
o Depth-wise: It explores deeper into the tree
or graph by following one path as far as
possible.
o Backtracking: After reaching a node with no
unvisited neighbors, it backtracks to the
most recent node with unexplored paths.
 BFS:
o Level-wise: It explores all nodes at a given
level before moving to the next level.
o Queue-based: Uses a queue to keep track
of the nodes to visit in level order.

3. Data Structure Used:

 DFS:
o Uses a stack (can be implemented with
recursion or explicitly with a stack) to
manage nodes to be visited next.
 BFS:
o Uses a queue to manage nodes in the order
they were discovered to visit the next node
level-by-level.

4. Space Complexity:

 DFS:
o Space complexity is O(d), where d is the
maximum depth of the tree or graph (the
deepest path). In the worst case (if the
graph is very deep), DFS can use a lot of
memory, especially for deep recursion.
 BFS:
o Space complexity is O(b^d), where b is the
branching factor (the maximum number of
children per node) and d is the depth. BFS
can require a large amount of memory since
it needs to store all nodes at the current
level before moving to the next level.

5. Time Complexity:

 DFS:
o Time complexity is O(V + E), where V is the
number of vertices (nodes) and E is the
 number of edges. DFS needs to explore all
vertices and edges in the worst case.
 BFS:
o Time complexity is also O(V + E). It explores
every vertex and every edge, making it
linear in terms of the graph size, similar to
DFS.

6. Completeness:

 DFS:
o Not complete in the case of infinite-depth
graphs or trees. DFS can get stuck exploring
an infinitely deep path and might never find
the solution even if it exists at a shallow
level.
 BFS:
o Complete in finite graphs or trees. BFS
guarantees that if a solution exists, it will be
found in the shallowest depth possible,
meaning it will explore all nodes level by
level until the goal is reached.

7. Optimality:

 DFS:
o Not optimal. DFS does not necessarily find
the shortest path to a goal, as it may
explore a long path first even if a shorter
path exists.
 BFS:
o Optimal (in unweighted graphs). BFS is
guaranteed to find the shortest path to the
goal in an unweighted graph or tree since it
explores all nodes at each depth level
before moving deeper.

8. Use Cases:

 DFS:
o Useful when:
 You want to explore all possible
solutions (e.g., solving a maze or
searching for all connected
components in a graph).
 Memory is a concern (especially
when the branching factor is large).
 You are looking for a path but not
necessarily the shortest path (e.g.,
puzzle-solving where any solution
will do).
  BFS:
o Useful when:
 You need the shortest path in an
unweighted graph or tree (e.g.,
finding the shortest route in a city
map).
 You need to explore nodes level by
level (e.g., social networks, word
ladders).
 You want to explore shallow
solutions before deep ones,
especially in finite graphs.

9. Example:

 DFS:
o Example in a tree:

sql
Copy code
DFS (Pre-order): Start at root
(1), go deep into each branch:
Path: 1 → 2 → 4 → 8 → backtrack →
5 → backtrack → 3 → 6 → 7

 BFS:
o Example in a tree:

sql
Copy code
BFS (Level-order): Visit all
nodes at the current level before
going deeper:
Path: 1 → 2 → 3 → 4 → 5 → 6 → 7 →
8

Summary of Differences:
Depth-First Search Breadth-First Search
Feature
(DFS) (BFS)

Depth-first (follows Breadth-first

Exploration one path as far as (explores nodes level
possible) by level)

Data Structure Stack (or recursion) Queue

O(b^d), where b is the

Space O(d), where d is
branching factor and
Complexity depth
d is depth

Time
O(V + E) O(V + E)
Complexity

Completeness Not complete (in Complete (in finite

 infinite spaces) spaces)

Optimal (finds the

Not optimal (can
Optimality shortest path in
find a long path)
unweighted graphs)

Finding any path,

Finding the shortest
solving puzzles,
Use Cases path, level-by-level
traversing deep
exploration
spaces

Backtracking Yes No

More memory Less memory

Memory
efficient for wide efficient, especially in
Efficiency
graphs wide graphs

5 Describe the key components of a search algorithm in the 8 L2

context of problem-solving agents.
In the context of problem-solving agents, a search
algorithm is a systematic method used to explore
possible states of a problem in order to find a path from
an initial state to a goal state. The goal is to find a
solution or an optimal solution efficiently. To do so,
search algorithms rely on several key components.
These components define how the search space is
explored and how decisions are made to reach the goal.

Key Components of a Search Algorithm:

1. State:
o A state represents a specific configuration
or condition of the problem at any point in
time. Each state can be thought of as a
node in the search space.
o The initial state is where the problem-
solving agent starts, and the goal state is
the desired outcome or solution.
o For example, in a maze, a state could
represent the current position of the agent
in the maze.

2. Initial State:
o The initial state is the starting point of the
problem-solving process. It represents the
configuration from which the search begins.
o The agent uses the initial state to begin
exploring the state space.
o Example: In a chess game, the initial state is
the starting arrangement of the pieces on
the board.
3. Goal State:
o The goal state is the target configuration
that the problem-solving agent seeks to
achieve. The search algorithm terminates
when the goal state is found.
o A well-defined problem has a clear goal
state or a goal condition that specifies when
a solution is found.
o Example: In a pathfinding problem, the goal
state is reaching a specific destination.

4. Actions:
o Actions are the set of possible moves or
operations that the agent can perform to
transition from one state to another.
o Each action changes the current state to a
new state by following the problem’s
defined rules.
o Example: In a navigation problem, actions
might include moving north, south, east, or
west to explore new locations.

5. Transition Model:
o The transition model defines how actions
transform the current state into a new
state. It describes the result of taking a
particular action from a given state.
o It is often represented as a function:
Result(s, a) returns the new state after
performing action a in state s.
o Example: In a puzzle, the transition model
specifies how swapping two pieces leads to
a new arrangement.

6. Path Cost:
o The path cost is a numerical value that
represents the total cost of a sequence of
actions from the initial state to the current
state. It allows the algorithm to evaluate
and compare different paths.
o The path cost might depend on factors like
distance, time, or any other resource. In
many algorithms, the goal is to minimize
this cost.
o Example: In a navigation problem, the path
cost might represent the total distance
traveled.

7. Search Space:
o The search space consists of all the possible
 states and actions that can be explored to
solve the problem. It represents the entire
set of configurations the agent could
encounter while searching for a solution.
o The size and structure of the search space
impact the efficiency of the search
algorithm. Larger or more complex search
spaces are harder to navigate and require
more efficient strategies.

8. Search Strategy:
o A search strategy is the algorithm or
method that dictates how the agent
explores the search space. Different
strategies define the order in which nodes
are visited and how paths are evaluated.
o Common strategies include:
 Depth-First Search (DFS): Explores
as deep as possible before
backtracking.
 Breadth-First Search (BFS): Explores
nodes level by level.
 Uniform-Cost Search: Expands the
least-cost node first.
 A Search*: Uses heuristics to guide
the search toward the goal
efficiently.
o The choice of strategy determines the
performance in terms of time, space,
completeness, and optimality.

9. Solution:
o A solution is a sequence of actions (or a
path) that leads from the initial state to the
goal state.
o The search algorithm terminates when a
solution is found. Some algorithms aim to
find the first solution (like DFS), while others
focus on finding the optimal (least-cost)
solution (like BFS or A*).

10. Heuristic Function (optional but important in

informed search):
o A heuristic function is an estimate of the
cost to reach the goal from a given state. It
helps guide the search process by
prioritizing states that appear to be closer
to the goal.
o Heuristics are often used in informed
search algorithms like A* to improve search
efficiency by focusing on the most
 promising paths.
o Example: In a navigation problem, the
heuristic might estimate the straight-line
distance to the destination.

11. Evaluation Function (for informed search):

o The evaluation function combines the path
cost and the heuristic to rank nodes in the
search. A common evaluation function used
in A* search is f(n) = g(n) + h(n),
where:
 g(n) is the actual cost to reach the
node n.
 h(n) is the estimated cost from n
to the goal (heuristic).
o This helps the agent balance between
exploring new paths and sticking to known
efficient paths.

Example: Pathfinding in a Grid

 State: The current position of the agent on the grid.

 Initial State: The starting position (e.g., top-left
corner).
 Goal State: The destination position (e.g., bottom-
right corner).
 Actions: Move up, down, left, or right.
 Transition Model: Moving in a direction results in a
new position.
 Path Cost: The total number of moves made
(assuming uniform cost).
 Search Space: All possible positions in the grid and
the transitions between them.
 Search Strategy: A* search with a heuristic based
on the Manhattan distance (sum of the absolute
differences of the coordinates) to the destination.
 Solution: The sequence of moves that takes the
agent from the start to the destination.

6 Summarize the concept of state space in relation to 5 L2

problem-solving agents.
The state space in relation to problem-solving agents
refers to the complete set of all possible states that can
exist within a problem environment. Each state
represents a unique configuration or situation that the
agent might encounter as it searches for a solution.

Key Aspects of State Space:

1. States:
o A state is a snapshot of the world or the
 problem at a particular moment. It
encapsulates all the necessary information
the agent needs to make decisions.
o Example: In a chess game, a state includes
the positions of all pieces on the board.

2. Initial State:
o The initial state is where the agent begins
its search. It defines the starting conditions
from which the agent explores the state
space.

3. Goal State:
o The goal state is the desired configuration
that represents the solution to the problem.
The agent's objective is to navigate the
state space to reach this state.

4. Actions and Transitions:

o Actions are the operations or moves the
agent can take to transition from one state
to another.
o The transition model defines how an action
changes one state into another, moving the
agent through the state space.

5. Search Space:
o The state space forms the search space for
the agent. The search space includes every
state and the connections (transitions)
between states that can be explored.

6. Path:
o A path is a sequence of states that the
agent transitions through as it explores the
state space. The path begins at the initial
state and ends at the goal state if a solution
is found.

Example:

In a maze-solving problem:

 The state space consists of all possible positions the

agent can occupy within the maze.
 The agent starts at the initial state (starting
position), and the goal is to reach the goal state
(exit of the maze).
 The agent uses actions like "move up," "move
down," "move left," or "move right" to transition
between states in the state space.
  The agent's task is to explore the state space
efficiently to find a path from the initial state to the
goal state.

Importance in Problem Solving:

The state space is fundamental to how problem-solving

agents approach tasks. By systematically searching or
exploring the state space, the agent identifies the best
sequence of actions (or path) to achieve the goal. The
size and complexity of the state space significantly
affect the performance of the search algorithm, making
the state space a central concept in problem-solving.