Get the basics of Machine learning…
Machine learning is the process of learning and training of the machines itself , in the same way as human learn in day to day life with the past experience. They basically help us to train and test the data on some parameter and helps us in finding out the accuracy and prediction of the trained model.
They are generally of 3 types-
1 ) Supervised learning
2) Unsupervised learning
3) Reinforcement learning
Before moving further ,please look at the given diagram so that you’ll get the basic idea of the structure of machine learning and its sub-parts.
- Supervised learning — In supervised learning for a given data-set , we are provided with the outputs or labels for the corresponding input , which makes it pretty easier for us to predict .
for example — Assume there’s a data-set of spam filtering in which we have been provided with messages and labelled column i.e whether that particular message is spam or ham(not -spam).
Ok , don’t worry if you weren’t able to understand , i am sharing the actual pic of how it looks.
Further moving on , Supervised learning is divided on 2 sub parts.
i ) Classification supervised learning
ii ) Regression supervised learning
Till now you might be familiar with the labels or output. Now in its sub-parts its just the type of the output .
i) Classification supervised learning- In this supervised learning the labels or output which will be provided in the data-set would be having discrete values i.e for every X input ,the corresponding Y output would be either 1 or 0 (depend upon data-set).
for example- In the spam data- set which i have taken above ,if you see the diagram again than you’ll get to see the labels as only spam or ham for any message , that is the example of Classification supervised learning.
The algorithms like logistic , SVM comes under this.
ii ). Regression supervised learning — In this supervised learning the labels or output which will be provided in the data-set would be continuous i.e for every X input ,the corresponding Y output would be different.
for example- the algorithms like linear regression comes under this .Here with the diagram you can better understand.
2. Unsupervised learning — In Unsupervised learning for a given data-set , we are not provided with the outputs or labels for the corresponding input , which makes it pretty difficult for us to predict .
For example — Well the only type of learning fall into this category is Clustering .
Clustering Unsupervised learning — In clustering , firstly we have to select the column or attribute on which we would be working and then visualize it .If the data- points are clustered then its good for you because you dont have to do much work then otherwise you have to see the data-point where it is lying.
The algorithm which is widely used for this type of dataset is K-Means Algorithm. I am sharing the link with you guys ,go visit there, you’ll definitely learn something.
https://www.youtube.com/watch?v=4R8nWDh-wA0
3) Reinforcement learning- In simple term here the machine or software agent perform certain activity and learns from itself when gone wrong from its past experience(you can relate to human) .
In the above example as you can see the agent is doing some action in the environment then the machine is interpreting the action ,if the action done is successful or right then the agent would be awarded otherwise it will just come back to its earlier position.
That’s it for now.If you like the blog do comment and help others by sharing.
-Mahesh Singh Dasila