Artificial intelligance

Definition of artificial intelligence:

Artificial intelligence is a branch of computer science concerned with the study and creation of computer system that exhibit so some from intelligence. Systems that learns now concepts and task system that can reason and draw useful conclusion about the world around us.

It is the science and engineering of making intelligent machine intelligent computer program. It is limited similar task of using computers to understanding human intelligence but artificial intelligence does not have confine itself to methods. In other words, we can say that artificial intelligence is the field of computer science which develop the ability to understand things in computer system like the human minds.

Artificial intelligence requires methods for encoring knowledge in computer system. Basic research in artificial intelligence had developed the variety of methods for this. These methods include predicate logic, production rules, semantic networks frames and script.

Intelligence means the ability to acquire, understanding and apply knowledge or the ability of exercise thoughts and reason.

·         Voice and speech recognition:

·         Speech understands is the recognition and understanding by computer of spoken language.

·         Robotics and sensory system:

·         Sensory system such as vision systems tactile system and signal processing system when combines with AI, define a broad categories of system generally called robotics. A robot is an electromagnetic device that can be programmed to perform manual task. An intelligent robot includes some kind of sensory such as cameras that collects information about the robot’s operations and its environment.

·         Computer vision: the basic objective of computer vision is to interpret scenarios rather than generate picture. For example, picture taken by satellite it may be sufficient to roughly identify reasons of crop damage. On other hand, robot vision system may be design precisely identify assemble component to being a command.

·         Neural computing:

·         A neural network is mathematical model of the way brain function. Neural networks are starting to have positive impact in many business disciplines.

·         Automatic programming:

·         The goal of automatic programming is to create special program that act as intelligent tools to assist programmers and fastly each phase of programming process

·         Genetic algorithm:

·         These are intelligent heuristic search methods that close a process that simulates evolution in the computer. Genetic algorithm has been applying to large scale combination mathematical programming problem such as large scale scheduling problems even the producing police sketches of criminals.

·         Fuzzy logic:

§  This extends the notation of logic beyond simple true false to allow or partial or continues truth in correct knowledge in precise reasoning are important aspects of expertise in apply common sense to decision making situation. One of the commercial applications of fuzzy logic was in producing superior antilock brakes.

·         Intelligent agent:

§  These are intelligent search methods that follow a process that simulates evolution in computer. For example, in a specific problem the solution of representing as chromosome which is generally contains the sequence of zero and one including the value of vector decision variables.

Problem description

Introduction of problem in AI:

AI problems are the problem that requires some kind of intelligence, computer conventional problem. AI problem a very broad spectrum they appear to have very little in common except that they are hard.

Before a solution can be found the important point is the problem must be very precisely defining such that the abstract problem should be converted into real workable states which are clearly understood then we apply that AI technique to solve the problems.

Problem solution systems:

Problem solving require following four things to be done. 
1. Problem description which consists of description initial situation and final situation

which is expected from the system and solution is accepted. 
2. Analysing the problem so that appropriate technique can be used to solve the problem. 
3. Isolation of knowledge and represent the task knowledge to solve the problem. 
4. Choosing the successful problem solving techniques and its applications to the specific

problem.

Problem characteristics: -

There are many methods to solve the AI problems for example state space search heuristic methods. These methods solve different AI problem but with limited capability that they solve the problem which belong to different areas such that each technique of problem solving each devoted to different class problems.

Choosing the most appropriate method is difficult and requires the implementation of multiple techniques to solve the problem. each problem is analyzing with following key attributes

1. Problem decomposability:

Can be broken into a set of independent small sub problems

2. Ignorable and undone step:

Can the problem solution method can ignore the steps or at least undone if method cannot prove the problem?

3. Predictability:

Can the problem outcome be universally certain or uncertain?

4. Absolute and relative solution:

Can the problem method provide the good solution compare to all other possible solution?

5. Solution is state or path:

Can the problem solution method provide the well define steps or path to provide the desired output?

6. Role of knowledge:

Is knowledge requiring solving the problem?

Absolute and relative solution:

On the basis of solution, the problem can be divided in to two types:

1. Any path problem or relative solution.

2. Best path problem or absolute solution.

1. In any path problem we have more than one path to successfully reach at the solution is

called relative solution problem.            

2. We have followed a particular path to solve the problem and that path is best for the

problem solving.

Production system:

Production system is applied to problem solving programs that must perform a wide range of searches. Production systems are symbolic artificial intelligence system. The difference between these two terms is only one of semantics. The symbolic AI system may not be restricted to the definition of production system but they cannot be different either.

Production systems are composed of three parts:

1. Global database

2. Production rules.

3. Control structure.

1. A global database is the system’s shortest or short term memory. These are collection of facts that are to be analyzed. A part of global database represents the current state of the system environment. In a game of chasse, the current state could represent all the position of the pieces on the chasse board.

2. Production rules are conditional if then branches. In the production system whenever a

condition in the system is satisfied, the system is allowed to execute the specific action or perform it which may be specified under the rule. If the rule is not fulfilled it may perform another action.

3. Control structure decides which production rules to use and in what sequence. Rules are picked up from the global database to used next in a production system algorithm.

Control strategies:

Search is one of the central issues in problem solving program or system in AI. Every AI program depends on a search procedure to perform its pre scribed functions. Problems are typically defined in terms of states and solution corresponds to the goal states.

Production systems provide a set of rules on the basis of which we can start searching from initial position to goal state. The production system can provide more than one rule for a state then the problem solving system required some kind of decision making ability or production or find out the right rule for that states. This ability or procedure is known as control strategies. Control strategies methods:

There are two methods

1. Breath first search (BFS)

 2. Depth first search (DFS)

1. Breath first search (BFS):

Breath first search is simple strategy in which the root node expended first, then all successors of the root node are expended next. Then their successor and so on. In general, all the nodes are expended on given depth in search tree before any nodes at next level are expended.

Breath first search can be implemented by queue structure where first in first out (FIFO) properties is maintained means the nodes that are visited first will be expended first.

2. Depth first search (DFS):

DFS is perfume by dividing downwards into a tree. It does this by generating a child node from most recently expended node than generating the child node child’s children until a goal is found or cut of depth point d is reached. An algorithm for DFS strategy can be implemented with stack. Where last in first out (LIFO) properties are maintain. Depth first places the newly generated at the head of the stack. So they will be chosen first.

Heuristic search technique:

The basic idea of heuristic search is that rather than try all possible search paths, you try and focus on paths that seem to be getting you near the goal state.

When all information then the initial state the operational and the goal state is available, the size of search space can usually be constraints the better the information available, more efficient the search methods process will be, such methods are known as informed search methods or heuristic search methods. Heuristic are often regarded as efficient they were incomplete for uncertain. this is not always true because some heuristic has a sound mathematical basis while some have precise knowledge with the high utility. Generate and test:

This method works in two modules.

1. The generator which creates possible solution.

2. The tester which evaluates each proposed solution either accepting or rejecting that

solution.

The process is terminated when one acceptable solution is found either accepting or rejecting that solution. The basic idea is to generate candidate solution in fact the solution to the problem. Candidate solution are trying by generating different numbers and testing them. This is known as blind search because there is no knowledge that is being used to directed the search and also search is unsystematic. This method is basically depth first search as complete solutions must be created before using testing. The algorithm is generating and test is given below:

Algorithm: generate and test:

Step 1. Generate a (potential goal) state:

- particular point in the problem space, or - a path from a start state

Step 2. Test if it is a good state:

- Stop if it is a good state - Otherwise go to step1