Essay: States and searching / Problem Solving As Search / State Spaces

C3) STATES AND SEARCHING
C3.1 Problem Solving As Search
Agent reasoning is about what an agent should do, the agent has a state-based model of the world with a certain objective to be reached. Agent reasoning is a flat (non-hierarchical) representation or a single level of a hierarchy representation. The agent can decide how to accomplish its objectives by looking in its representation of the world state space for a path to advance its current state to a goal state. It can search for a series of actions that will accomplish its goal before it has to proceed to take actions in the real world.
Search highlighted much of artificial intelligence. When an agent is presented with a problem, it is usually given only an explanation that lets it acknowledge a solution, not an algorithm to solve it. It is obligatory for it to search for a solution. The existence of NP-complete problems, which means the solution founded but no efficient methods for finding them, indicates that searching plays a vital role in solving problems.
C3.2 State Spaces
One general construction of artificial intelligent is in state space. A state consists of all of the information needed to forecast the implications of an action and to decide if it is a goal state. State-space searching is supposedly:
• The agent has ideal knowledge of the state space and can observe what state it is in.
• The agent has a series of actions that are known for its deterministic effects;
• The agent wants to achieve one of these goal states, and the agent can recognize a goal state; and
• a solution is a series of actions that will advance the agent from its current state to a objective state.
C3.3 Graph Searching
The general mechanism of searching is presented in terms of searching for paths which leads to a goal path in directed graphs. To find the solution for a problem, determining the underlying search space first and then appeal a search algorithm to that search space. Many problem-solving tasks can be converted into the problem of searching for a path in a graph. Searching in graphs supplies a suitable level of illustration within which to study simple problem solving of a particular domain.
A (directed) graph contains a set of nodes and directed arcs leading to goal node. The abstraction is needed as there are many ways to illustrate a problem as a graph.
Problem Solving by Graph Searching
C3.4 A Generic Searching Algorithm
Generic searching algorithm is the searching of a solution path in a graph. The algorithm stands alone of any certain search strategy and any certain graph.
Generic graph searching algorithm
C3.5 Uninformed Search Strategies
A problem determines the graph and the goal but not which path to select from the start node to the goal node. But, a search strategy determines which paths are selected from the start node to the goal node. Different strategies are acquired by altering how the selection of paths in between the partition is implemented.
There are 3 uninformed search strategies that do not consider the location of the goal. They are depth-first search, breadth-first search and Lowest-Cost-First Search. These search strategies ignore which path they are taking until they find a goal and report success.
The order nodes are expanded in depth-first search
The order in which nodes are expanded in breadth-first search
C3.6 Heuristic Search
The heuristic function is to inform the search about the path leading to a goal. It provides an way to forecast which neck of a node will lead to a goal. A standard way to obtain a heuristic function is to find a solution to a simpler problem.
• A* search is a merger of lowest-cost-first and best-first searches that interprets both path cost and heuristic information in its decision of which path to expand to lead to its goal.
•
Strategy Selection from Frontier Halts? Space
Depth-first Last node added No Linear
Breadth-first First node added Yes Exponential
Best-first Globally minimal h(p) No Exponential
Lowest-cost-first Minimal cost(p) Yes Exponential
A* Minimal cost(p)+h(p) Yes Exponential
Summary of Search Strategies
C3.7 More Sophisticated Search
A number of improvements can be made to the foregoing strategies. First, the two methods that are appropriate to be implement when there are cycles in the graph; one checks precision for cycles, whereas the other technique checks for several paths to a node. Next, iterative deepening and depth-first branch-and-bound searches, are general methods that are sure to search for the solution (even an optimal solution), like breadth-first search or A* search, but using the space pros of depth-first search. Then problem-reduction method is to simplify a search problem into various smaller search problems, easier to solve each problems. Finally dynamic programming is particularly used for searching of path and for building heuristic functions.