Machine learning basic concept for beginners.
In this blog, we will study some basics of machine learning, even if you are a beginner don't worry you will understand all the basics in this blog.
 |
| Machine learning basics. |
Table of contents
- History of machine learning.
- Meaning of machine learning.
- Definition of machine learning.
- Basic methods of machine learning.
- Supervised learning.
- Unsupervised learning.
- Reinforcement learning.
History of machine learning.
The term machine learning popularized by an American computer scientist Arthur Samuel in 1959. He was a pioneer of machine learning. He was born in 1901 in emporia, Kansas. He was best known for checkers (GAME) playing program.
Arthur Lee Samuel believed that teaching computers to play games like checkers is very developing tactics appropriate to general problems.
In 1949 he moved to international business machines corporation (IBM).
He developed the checkers playing programs just for attraction because at that time he did not have money for his ongoing project ILLIAC (Illinois Automatic Computer). This project had a goal of developing supercomputers.
. The meaning of machine learning.
 |
| Meaning of machine learning. |
Before understanding what is meant by machine learning, we need to understand what, is the term learning means, learning is a process of understanding, gaining or obtaining new knowledge, developing new skills, improving or understanding the behaviours.
Learning is an aspect of cognition, and cognition is a mental process of obtaining knowledge, understanding the thoughts, understanding how we think.
We as a human learn through our previous experiences.
Now the same process applies to machine learning, the machine or a system learns the same as humans. According to Josh Kaufman every human needs 20 hours to learn any new skill and requires 10,000 hours to master a skill.
. The definition of machine learning.
Machine learning is a field or application of artificial intelligence.
To know about artificial intelligence click here.
The system learns things automatically and improvs through its own previous or self - experience.
The three basic methods of machine learning.
1) supervised learning.
2) unsupervised learning.
3) reinforcement learning.
1) supervised learning.
In simple words, supervised learning is a technique in which we need to train a system or algorithms to accomplish or to complete a task. In supervised learning, We need to put the input in a system to get the required output.
Example: exit polls in elections. We take feedback or data from all age groups of people (young, old). This data or feedback is input know by using this input data We predict the output.
Supervised learning has two types.
A) linear regression supervised.
B) classification supervised.
A) linear regression.
Linear regression is an algorithm of machine learning which can predict the dependent variable (y) using a given independent variable (x). In linear regression, if we put two input values in a system we will get two output values.
B) classification regression.
It is another type of supervised learning. In linear regression we need to put the exact amount of input values to get the exact amount of outputs.
But, classification regression is quite different here if we put two independent variables or two input values we will get more than two outputs, because here it can predict more outcomes by observing by using his previous input values.
2) unsupervised learning.
Unsupervised learning is a self - learning process. Here the system learns through experience using the previous input data.
Example : a new born baby will don't know anything but the baby will learn the things growing day by day, by using his previous day experience.
Types of unsupervised learning.
- clustering.
- association.
- clustering.
In clustering the data will be divided into groups based on the similarities. It is a process of arranging it organising the objects based on the similarities.
Types of clustering.
four types of clustering algorithms are:
a) exclusive.
b) hierarchical.
c) overlapping.
d) probabilistic.
- association.
It is an unsupervised learning method that attempts to find the connection between different types of entities.
3) reinforcement learning.
It is the process of training a model to make decisions. In this method we need to train the algorithms using reward and penalty or punishment. The goal is to reach maximum rewards.
In reinforcement learning the algorithms will avoid punishment, so if it gets rewards it will learn and repeat the process to get more rewards, but if it's penalized, it will not repeat those mistakes again.
like humans, we don't want to repeat the same mistakes where we failed to gain the rewards, to earn rewards we will not repeat those mistakes again and again.
Types of reinforcement learning.
a. postive reinforcement.
b. negative reinforcement.
c. positive punishment .
d. negative punishment.
a. Positive reinforcement
Positive reinforcement is an incorporation of reward to a desired behavior. The reward system will improve the behavior to do something good or the reward system will make the behavior to occur again.
Example
A treat to a child for completing this homework. This reward (treat) will make the child complete his homework.
b. Negative reinforcement
In positive reinforcement we saw an addition of reward for improving a desired behavior. Negative reinforcement is opposite to positive reinforcement.
Negative reinforcement is removing or avoiding a behavior for the best outcome.
Example
Spanking a child to wash his hands before eating.
c. Positive punishment
positive punishment is something that reduces undesired behavior. This punishment helps to overcome the undesired or unwanted behavior.
Example
Embarrassing a student in front of the whole class for his low marks in a subject will make him get good marks in the next attempt.
d. Negative punishment
Negative punishment is similar to positive punishment. Negative punishment is seizing or removing a favourite item from a person or list.
Example
Seizing a mobile phone from a child or kid because he is not completing his Homework.
Summary
In this blog, we have studied the basics of machine learning, the meaning of machine learning. Now, we know when the term machine learning is popularized, we also know the history of machine learning. We have learnt the three basic types of machine learning.
The supervised learning, unsupervised learning and reinforcement learning.We studied the four types of reinforcement learning: positive reinforcement, negative reinforcement, positive punishment, negative punishment.
0 Comments