LEARNING

Learning

Learning is a process by which a system acquires knowledge.

Learning is the activity that enables the system to do the same
task more efficiently and more effectively next time.
Learning is the key to improve a system’s performance.
Approaches towards Learning

Skill Refinement
Knowledge Acquisition
Taking Advice
Problem Solving
Induction
Discovery
Analogy
 Learning
Learning is an incessant(continuous) process.
It is done by viewing ,listening , interacting
with others , studying etc. and also, by
experience.
It helps us in:
Acquiring new knowledge ;
Refining and updating the knowledge ;
Provides us power to handle problems.
 Contd..
Basically, learning is a unique skill possessed by
us that infuses in us power to reason, capability
to deal with new situation and in one sense ,
enables us to act intelligently.
Same way, if a machine has to be intelligent , it
needs to possess skill of learning.
Thus, making a machine intelligent would require
making it to learn as a machine cannot be called
intelligent , if it does not have power of learning.
 Contd..
Learning is important for practical
applications of Artificial Intelligence.
Learning can be defined as, “ Any changes in a
system that allows it to perform better the
second time on repetition of the same task or
on another task drawn from the same
population”.
 operatio
n
output

Input program execution Result

Use acquired
knowledge for
future input Store the result
programs for future use

Knowledge
acquired(for
future use )

A general learning
system
 Contd..
In the above drawn diagram, the input program
is any general program executed for solution of a
problem.
The results are reported as output, but results
are also stored back to acquire knowledge for
future use.
It means, in case similar problem is required to
be solved next time, the results can automatically
be taken from previously acquired knowledge.
 How can we learn?
Skill refinement
one can learn by practicing, e.g. playing the piano.
refers to the situation of improving the skill by performing
same task again and again.
Knowledge acquisition
one can learn by experience and by storing the
experience in a knowledge base. One basic example of
this type is Rote(habitual repetition) memorized learning.
Taking advice
Similar to rote learning although the knowledge that is
input may need to be transformed (or
operationalized) in order to be used effectively.
 Contd..
Problem Solving
if a person solves a problem he may learn from this
experience. The next time he sees a similar problem he
can solve it more efficiently. This does not usually involve
gathering new knowledge but may involve reorganization
of data or remembering how to achieve the solution.
Induction
Humans often classify things in the world without knowing
explicit rules. Usually involves a teacher or trainer to aid
the classification.
Discovery
Here one learns knowledge without the aid of a teacher.
Analogy
If a system can recognize similarities in information
already stored then it may be able to transfer some
knowledge to improve the solution of the task at hand.
 Major paradigms of machine learning

Rote learning – One-to-one mapping from inputs to stored representation.

“Learning by memorization.” Association-based storage and retrieval.
Induction – Use specific examples to reach general conclusions.
Analogy – Determine correspondence between two different
representations.
Discovery – Unsupervised, specific goal not given.
Clustering – Unsupervised identification of natural groups in
data.
Genetic algorithms – “Evolutionary” search techniques, based on an
analogy to “survival of the fittest”.
Reinforcement – Feedback (positive or negative reward) given at the end
of a sequence of steps.
 1. Learning from examples
This type of learning is also known as Learning by
Induction.
Induction means “the inferring of general laws from
particular instances”.
In this, machines learns to perform similar actions in
similar situations.
Based on these examples, we formulate certain rules
like: e.g. when there are clouds in the sky, it rains–
means clouds bring rain.
Learning by examples (contd..)
Inductive learning means, generalization of
knowledge gathered from real world
problems and use of the same for solving
other similar problems.
Thus inductive learning or learning from
examples is a method which generalizes the
rules of past.
The action may be internal action.
 Learning by examples (contd..)
Generally, in this , the L.H.S is said to be the concept and
R.H.S , the rule.
The Inductive learning can be rep. as:
Situation 1 ------ action 1
Situation 2 ------ action 2
Situation n ------ action n
Generally , in , AI applications, we create a rule base of “IF-
THEN” type of structure which is “Situation-Action”
sequence.
The Induction learning method, while solving a problem,
uses this “situation-action” sequence and in future, given
the similar situation, automatically takes the similar action.
 Learning from Examples
Inductive Learning or Concept Learning

It is concerned with finding a hypothesis from groups of positive and

negative examples.

positive examples + negative examples + background knowledge => hypothesis

It is a system that tries to induce a general rule based on observed instance.

It is a kind of supervised learning, where examples are provided with

classifications.
 The inductive learning problem

Extrapolate from a given set of examples to make accurate predictions about

future examples.
Supervised versus unsupervised learning
Learn an unknown function f(X) = Y, where X is an input example and Y is
the desired output.
Supervised learning implies we are given a training set of (X,
Y) pairs by a “teacher”
Unsupervised learning means we are only given the Xs and some
(ultimate) feedback function on our performance.
Concept learning or classification
Given a set of examples of some concept/class/category, determine if a
given example is an instance of the concept or not.
If it is an instance, we call it a positive example.
If it is not, it is called a negative example.
Or we can make a probabilistic prediction (e.g., using a Bayesian net)
 The Inductive Learning Problem

Given: Training examples (x; f(x)) for some unknown function f

Find: A good approximation to f.

Example Applications:-
Handwriting Recognition
x: Data from pen motion.
f(x): Letter of the alphabet.
Disease diagnosis
x: Properties of patient (symptoms, lab tests)
f(x): Disease (or maybe, recommended therapy)
Face recognition
x: Bitmap picture of person's face
f(x): Name of the person.
Spam Detection
x: Email message
f(x): Spam or not spam.
 2. Discovery as learning.
It is a restricted form of learning that acquires
knowledge without the help of a teacher.
It has three categories:-
Theory Driven Discovery
Data Driven Discovery
Clustering
 3. Learning by Analogy.
Analogy (resemblance) involves a complicated mapping
between what might appear to be two dissimilar concepts. It
is a powerful inference tool.
Act cleverly like a fox.
He is just like a Lion.
Humans quickly recognize the abstractions involved and
understand the meaning.
There are two methods of analogical problem solving studied
in AI.
Transformational Analogy
Derivational Analogy
4. Explanation based Learning.
It is also called situation based learning and is
a kind of analogical learning.
The explanation based learning is knowledge
intensive approach (more emphasis is given
on knowledge).
There are certain applications where instead
of data, knowledge plays more important
role, e.g. chess playing, medical diagnosis.
 4. Explanation based Learning.
(contd..)
In EBL, the learned knowledge is valid knowledge which is
obtained by performing a deductive reasoning process on a
set of facts.
An EBL accepts 4 kinds of input:
A training example
what the learning sees in the world.
A goal concept
a high level description of what the program is supposed
to learn.
An operational criterion
a description of which concepts are usable.
A domain theory
a set of rules that describe relationships between objects
and actions in a domain.
 5. Genetic algorithms.
It is an example of biological learning.
It is based on analyzing how the evolution of
human beings took place.
Genetic algorithms are based on ideas that
are analogous in some ways to chromosome
crossover, gene mutation , fitness and natural
selection.
 Genetic algorithms (contd...)
Genetic algorithms are methods of breeding
computer programs and solutions to optimization or
search problems by means of simulated evolution
loosely based on natural selection , crossover and
mutation are repeatedly applied to a population of
binary strings, which represent potential solutions.
A genetic algo requires 2 things to be defined:
1. A genetic rep. of the solution domain.
2. A fitness function to evaluate the solution domain.