UNIT - II
Topic: Search with Partial Information (Heuristic Search)
Part A: Introduction
✅ What is Heuristic Search?
Heuristic Search is an informed search strategy that uses additional knowledge (called a
heuristic) to make better decisions during search.
Unlike uninformed (blind) search methods like BFS and DFS, which explore the search
space blindly, heuristic search tries to guess which path is more promising by using a
heuristic function.
Part B: Key Terms in Heuristic Search
Term Description
A rule or estimate that helps in decision-making (a guess of how
Heuristic
close we are to the goal)
Heuristic Function
Estimates the cost from the current node n to the goal
(h(n))
Evaluation Function Function used to rank which node to expand next
✅ Example of a Heuristic
In a map-based search, the straight-line distance (Euclidean distance) from a city to
the destination is a heuristic.
It’s not the actual road distance but helps estimate which city is closer to the goal.
Part C: Need for Heuristic Search
Uninformed search algorithms explore all possible paths, which is slow and memory-
intensive.
Heuristic search focuses only on promising paths, reducing time and space
requirements.
It is useful when:
o The search space is large
o The goal is far
o The agent has partial information
Part D: Types of Heuristic Search Algorithms
✅ 1. Greedy Best-First Search
Selects the node that appears to be closest to the goal using h(n) only.
Feature Description
Strategy "Always choose the nearest-looking node"
Function used f(n) = h(n)
Optimal? ❌ No
Complete? ❌ No (may get stuck in loops)
Example:
In a city map, it always chooses the city closest to the destination, even if it leads to a dead
end.
✅ 2. A Search Algorithm*
Combines both path cost so far (g(n)) and estimated cost to goal (h(n)).
Feature Description
Strategy Balance between shortest path and nearest goal
Function used f(n) = g(n) + h(n)
Optimal? ✅ Yes (if h(n) is admissible)
Complete? ✅ Yes
Example:
Finds the shortest and best route to the goal by considering both the distance covered and
the estimated distance left.
✅ 3. Hill Climbing Search
Chooses the neighbor with the best heuristic value and keeps climbing.
Feature Description
Strategy "Keep going up (towards better states)"
Feature Description
Function used Only h(n)
Optimal? ❌ No
Complete? ❌ No (may get stuck on local maxima)
Problem:
May get stuck at a local optimum, thinking it’s the best when it’s not.
Example:
Climbing a hill in the fog based on where the slope feels steepest — might reach a small
peak, not the highest one.
✅ 4. AO* Algorithm (And-Or Graph Search)
Used in problem reduction where tasks are broken into sub-tasks (AND nodes and OR
nodes).
Feature Description
Usage For non-linear, hierarchical problems
Graph Type AND-OR Graphs
Optimal? ✅ Yes (with admissible heuristic)
Application Used in expert systems, automated planning
Part E: Comparison Table
Uses Uses Path
Algorithm Complete Optimal Example Use Case
Heuristic? Cost?
Greedy Best-
✅ h(n) ❌ ❌ ❌ Navigation systems
First
A* Search ✅ h(n) ✅ g(n) ✅ ✅ Map routing, games
Optimization
Hill Climbing ✅ h(n) ❌ ❌ ❌
problems
Problem
AO* Search ✅ ✅ ✅ ✅
decomposition
Part F: Advantages and Disadvantages
Advantages Disadvantages
✅ Faster than uninformed search ❌ Heuristic must be well-designed
✅ Saves memory and time ❌ May not find the shortest path if heuristic is incorrect
✅ Finds solutions in large spaces ❌ Can get stuck in local minima (in Hill Climbing)
📝 Summary
Heuristic search uses extra knowledge to find solutions more efficiently.
A heuristic is a rule-of-thumb that estimates the closeness to goal.
Popular heuristic algorithms include:
o Greedy Search (fast, not optimal)
o A* Search (optimal and complete)
o Hill Climbing (simple but risky)
o AO* (used for complex task reduction)
Choosing a good heuristic function is key to successful problem solving in AI.
"Heuristic search is like using your brain while exploring — not just walking blindly!"
