ai search techniques

15
Search Techniques in AI Omar Adnan Isaid

Upload: omar-isaid

Post on 13-Aug-2015

68 views

Category:

Technology


0 download

TRANSCRIPT

Search Techniques in AIOmar Adnan Isaid

What is search ?

• Sequence of steps used to solve a problem

• Types of search –Informed search (blind)–Uninformed search (heuristic)

Why we need heuristic

• The data or the problem statement is not clear enough.

• exponential explosion ;the solution available but there is no practical time to search for the solution.

Types of Problems

• Path finding problem is the problem of finding the path between the initial state and the goal state.

• The constraint satisfaction problems (CSPs) based on choosing a specified subset of objects from a certain set of objects, that its constraints and limitations satisfy certain conditions.

• two player game with perfect information, where each player can see the movements and events of the other player

Depth limited search

• The Depth limited search is designed to avoid the failure of Depth-First search (DFS) .Where, the (DFS) can stuck in cycles or infinite search space

Depth Limited Search

The Backtracking search techniques

• Backtracking is an algorithm suitable for constraint satisfaction problems (CSPs) : TSP,N*N Queens and others.

• a modified Depth First Search (DFS) algorithm• Uses the promising function to cut off the not

promising branches

Backtracking Algorithm

Intelligent Agent

The (Russell et al, 2009) defines the agent “anything can perceive its environment through sensors and actions upon through actuators”.

Adversarial Search(Russell,2009)

• Arises in multi agent hostile environment • Attractive and challenging for complex and

creative heuristic algorithms

Graph Tree , tic tac toe 9!=362880 states

Minimax algorithm

• Movement of the opponent is a major challenge in two player game

• Max player plays to win the game • Min player plays to prevent Max from winning

the game• Drawback : expects each branch is a promising

branch

Minimax algorithm Max blue , Min red

Minimax without alpha procedure

Minimax without alpha procedure