1. LEARNING

2. WHAT IS LEARNING 
The artificial intelligence is that machine that cannot be called intelligence until they are able to learn to do new things and to adapt to new question and new situation rather than simply doing as they are told to do. The ability to adopt new surrounding and to solve new problems is an important characteristic of intelligent entities.
The learning can accommodate knowledge acquisition. Knowledge is generally acquired through experience.
Another way we learn is through taking advises from other. People also learn through their own problem - solving experience. If we solve the problem first time. Then solving the same problem next time is efficient because we learn some thing during solving first time.
We learn to classify things in the word without being given explicit rule. Learning from the examples usually involves a teacher who helps us classify things by correcting us when we are wrong. Hence learning is itself a problem – solving process.

3. ROTE LEARNING
Rote learning is most basic learning activities when the computer stores the date it is performing rudimentary form of learning. Hence the act of storage allows the program to perform better in the future. Also in the case of data caching we store computed value then we do not recomputed again when the computation is more expensive then this strategy can save the significant amount of time. Hence caching has been used in AI program to produce some surprising performance improvements. Such caching is known as Rote learning.

4. Rote learning includes the capabilities
Organized storage of information: - In order to improve the performance and speed up to use the stored value than it would be to recomputed it. Then there must be a special technique that accesses the stored value quickly.
Generalization: - Here the number of distinct object that stores are very large . So that to keep the number of stored object manageable level some kind of generalization technique is necessary.

5. LEARNING BY TAKING ADVICE
In the computer when the programmer writes a series of instruction into a computer a rudimentary kind of learning is taking place. Hence the programmer is sort of teacher and computer is sort of student. Hence being programmed the computer is now able to do something it previously could not do.
People process advice in an analogous way. The different type of advice takes a particular
on different situation . Computer program might make use of the advice by adjusting its static evaluation function to include a factor depending on the other control. If we have design a data structure for playing any game then first we rule out all the advise before playing the game. Hence human user first translates the advice then plays the game.

6. LEARNING IN PROBLEM SOLVING
Learning in problem takes various technique to improve the performance. Like that problem solver solve the problem by taking advice from some one else or teacher.
Learning by parameter adjustment: - The most important question in the design of a learning program based on parameter adjustment. When the value of parameter increased and when the value of parameter decreased. The second question is how much should the value be changed. Hence the answer to the first question is that value of parameter that predicted the final outcome accurately should be increased while the value of parameter of poor predictors should be decreased. In designing the program we have to know a priority how much weight should be attached to each feature being used. The solution of this we estimates the weight of problem through solving.

7. LEARNING IN PROBLEM SOLVING(cont..)
LEARNING WITH MACRO_ OPERATORS: - Macro operators is consist of sequence of action perform in semantic way. A macro is just like a regular operator except that it consisting of sequence of actions not just a single one. A macro operation is preconditions are initial conditions of the problem just solved and post conditions correspond to the goal just achieved. Suppose we are given an initial block world situation in which ON(C, B) and ON (A, TABLE) are both true. The strips can achieve the goal ON (A, B) by dividing a plan with the four step UNSTACK(C, B), PUDOWN (e), PICKUP (A), STACK (A, B). Hence the macro with prediction ON (C, B) ON (A, TABLE) and post condition ON(C, TABLE) ON (A, B). The body of macro consists of four steps just mentioned. Hence for solving problem it is better to define certain macro operation in the problem.

8. LEARNING IN PROBLEM SOLVING(cont..)
LEARNING BY CHUNKING: - Chunking is the process similar to macro operators. The idea of chunking comes from psychological literature on memory and problem solving. It computation basis is in production systems. So that solving the problem we have to define the number of production in the memory depending upon the problem we called chunk from memory and solve the problem

9. LEARNING IN PROBLEM SOLVING(cont..)
THE UTILITY PROBLEM: - PRODIGY is a general-purpose problem solving system that incorporates several different learning mechanisms. A good deal of the learning in PRODIGY is directed at automatically constructing a set of control rule to improve the search in the particular domain. PRODIGY can acquire control rules in the number of ways
Through hand coding by programmer.
Through a static analysis of the domain operators.
Through looking at traces of its own problem solving behavior.
PRODIGY learns control rules from its experience. The major work in PRODIGY was the identification of utility problem in the learning systems. PRODIGY maintains a utility measure for each control rule.

10. LEARNING FROM EXAMPLES: INDUCTION
Classification is the process of assigning to a particular input the name of a class to which it belongs . Classification is an important component of many problem-solving task. Before the classification can be done the class it will use must be defined. This can be done in the way
Isolate a set of features that are relevant to the task domain. Define each class by a weighted sum be of values of these features. Each class is then defined by a scoring function that looks very similar to the scoring functions often used in other situation.
Isolate a set of feature that is relevant to the task domain. Define each class as a structure composed of those features.

11. WINSTON’ S LEARNING PROGRAM
The basic approach that Winston’ s program took the problem of concept formation can be described as:
Begin with structural description of one known instance of the concept call that description the concept definition.
Examine descriptions of other known instances of the concept. Generalize the definition to include them.
Examine descriptions of near misses of the concept. Restrict the definition to exclude these

12. WINSTON’ S LEARNING PROGRAM(cont..)
This process function has the same way as any other matching process. Such as one to determine whether a given production rule can be applied to a particular state. Because the differences as well as similarities must be found and the procedure must perform not literal but also approximate matching. The output of the comparison procedure is a skeleton structure describing the commonalities between the two input structures.

13. EXPLANATIONS BASED LEARNING
The explanations based learning system attempt to learn from a single example ‘x’, why x is an example of the target concept. The explanations are then generalized and system’s performance is improved through the availability of this knowledge. Hence explanation-based-learning we have to consider following such as:

14. EXPLANATIONS BASED LEARNING(cont..)
A TRAINING EXAMPLE: - What the learning program ‘sees’ in the world.
A GOAL CONCEPT:- A high level description of what the program is supported to learn.
AN OPERATIONALITY CRITERIA: - A description of which concept is usable.
A DOMAIN THEORY: - A set of rule that describe relationship between objects and action in a domain.
Hence from this explanations- based –learning computes generalizations of the training example that is sufficient to describe the goal concept and also satisfy the operational criterion.

15. EXPLANATIONS BASED LEARNING(cont..)
The training example is familiar input. The goal concept is not operational. The domain theory is rule of the solving the problem. The EBL is faced with the task of characterizing the entire range of objects that satisfy the goal concept. Most of these objects will never be encounter in the real world . There is evidence that human do not learn with very primitive relations. Instead they create incomplete and inconsistent domain theories. The difference between EBL and other learning method is that EBL programs are built to take advantage of domain knowledge. Since learning is just another kind of problem solving.

16. What is GA
A genetic algorithm (or GA) is a search technique used in computing to find true or approximate solutions to optimization and search problems.
Genetic algorithms are categorized as global search heuristics.
Genetic algorithms are a particular class of evolutionary algorithms that use techniques inspired by evolutionary biology such as inheritance, mutation, selection, and crossover (also called recombination).

17. What is GA
Genetic algorithms are implemented as a computer simulation in which a population of abstract representations (called chromosomes or the genotype or the genome) of candidate solutions (called individuals, creatures, or phenotypes) to an optimization problem evolves toward better solutions.
Traditionally, solutions are represented in binary as strings of 0s and 1s, but other encodings are also possible.

18. What is GA (Cont..)
The evolution usually starts from a population of randomly generated individuals and happens in generations.
In each generation, the fitness of every individual in the population is evaluated, multiple individuals are selected from the current population (based on their fitness), and modified (recombined and possibly mutated) to form a new population.

19. What is GA (Cont..)
The new population is then used in the next iteration of the algorithm.
Commonly, the algorithm terminates when either a maximum number of generations has been produced, or a satisfactory fitness level has been reached for the population.
If the algorithm has terminated due to a maximum number of generations, a satisfactory solution may or may not have been reached.

20. Key terms Individual - Any possible solution
Population - Group of all individuals
Search Space - All possible solutions to the problem
Chromosome - Blueprint for an individual
Trait - Possible aspect (features) of an individual
Allele - Possible settings of trait (black, blond, etc.)
Locus - The position of a gene on the chromosome
Genome - Collection of all chromosomes for an individual

21. Chromosome, Genes and Genomes

22. Genotype and Phenotype
– Particular set of genes in a genome
Phenotype:
– Physical characteristic of the genotype (smart, beautiful, healthy, etc.)