How Machine Learning (ML) Work

How Machine Learning (ML) Work

In this post, We will learn how machine learning work.

What is machine learning (ML) ?

Machine learning is a study of model that compute use to progressively improve their performance on a specific task.

What is Machine learning models ?

Machine learning model contains a set of instruction or mathematical formula or in simple word it contain a piece of code in which we provide some data as input and model return some output. We learn more about model and how use a machine leaning model latter.

How Machine Learning (ML) Work  ?

This part we will understand with an example. Suppose want to have a simple application which will take a picture and banana or apple and application will tell us either it is banana or apple or picture is neither banana nor apple.

To solve this problem first we provide a set of banana picture to our model and we will tell our model that this is a banana picture similarly we will provide a set of apple picture to our model and we will tell our model that this is apple picture. 

From these picture our mode our model will extract some properties of banana and apple like

Banana

• Shape is cylindrical 
• Colour is yellow

Apple

• Shape is spherical
• Colour is red  

Above two properties are only for understanding this example but in real there would be more properties. So now our model know how a banana and an apple look like. So now if we provide another picture to our model it will calculate the properties of this newly image and match its properties with already calculated properties and if matching is more close to banana means our new image is banana and if matching is more close to apple means our new image is apple.

Training and Testing of Machine Model

In training we teach our model how to behave, like in above example we train our model by providing initial set image to model and its name (Banana or Apple). In testing we provide input and our model predict out put, like after training our model had properties of banana and apple so in testing we provide another image as input to model and this time we do not provide information to model that image is of banana or apple instead our model calculate properties and predict output.

Please write comments if you find anything incorrect, or you want to share more information about the topic discussed above.