PROBLEM-SOLVING AGENT
Problem-Solving Agents
• When the correct action is not immediately
obvious, an agent must plan ahead by
considering a sequence of actions that lead to
a goal state.
• Such an agent is called a problem-solving
agent.
• The computational process it undertakes is
called search.
2. Representation in Problem-Solving
• Atomic Representation:
– States of the world are treated as indivisible wholes.
– No internal structure of the state is visible to the
search algorithms.
• Factored/Structured Representation: Used by
planning agents (discussed in advanced
chapters).
• States are represented with multiple variables
and relationships.
3. Types of Search Algorithms
• Uninformed Search (Blind Search):
– No knowledge of the goal distance is available.
– Examples: Breadth-First Search (BFS), Depth-First
Search (DFS), Uniform Cost Search (UCS), and
Iterative Deepening Search (IDS).
• Informed Search (Heuristic Search):
– Uses knowledge (heuristics) to estimate the
distance to the goal.
– Examples: Greedy Best-First Search, A Algorithm.*
4. Characteristics of the Simplest
Search Environments
• Episodic: The agent makes decisions one step at a time,
without considering past actions.
• Single Agent: Only one agent is involved in decisionmaking.
• Fully Observable: The agent has complete knowledge of
the environment.
• Deterministic: The outcome of an action is always
predictable.
• Static: The environment does not change while the agent is
thinking.
• Discrete: The state space consists of distinct states.
• Known: The agent has prior knowledge of the rules of the
environment.
5. Complexity of Search Algorithms
• The performance of search algorithms is
analyzed using asymptotic complexity (Big-O
notation).
• O(n) notation describes the efficiency of an
algorithm based on the input size.
Problem-Solving Process for an Agent
in Romania
• An intelligent agent on a touring vacation in
Romania faces a complex decision-making
problem.
• It must decide the best route to reach
Bucharest while considering multiple factors
such as sightseeing, language learning, and
nightlife.
• If the agent has no knowledge of the geography, it
may have to pick a road randomly—a situation
discussed in uninformed search ..
• However, if the agent has a map and knowledge
about the world, it can follow a structured problemsolving approach:
Four-Phase Problem-Solving Process
1. Define the Problem:
• Identify the initial state (Arad).
• Define the goal state (Bucharest).
• Determine possible actions (choosing between
roads to Sibiu, Timisoara, or Zerind).
2 .Formulate a Search Strategy:
• If the agent has a map, it can analyze possible
routes.
• The agent can use search algorithms like
Breadth-First Search (BFS), Depth-First Search
(DFS), or A Algorithm* to find the best path.
3.Execute the Chosen Path:
• The agent follows the selected route step by
step.
• It updates its knowledge based on
observations (e.g., traffic conditions,
roadblocks).
4.Arrive at the Goal and Evaluate:
• The agent reaches Bucharest successfully.
• It evaluates if the chosen path was efficient.