What is machine learning? – Definition

Machine learning: If you are a person who cares about the word called New technology then you must hear about the term called machine learning. and if you seeking the right definition of machine learning then you are in the right place.

So today am going to answer a very big and famous question that is what is machine learning and how it differs from other areas like data science and artificial intelligence?

What you are going to learn

1. What is machine learning?
2. Types of machine learning

What is Machine learning? definition

Machine learning is a branch of computer science that deals with different algorithms in different conditions. using these algorithms machines can perform a specific task. we can also use search patterns for performance.

If I say it in simple words then ML is nothing but how we can train a machine so it can learn and perform tasks on its own. 

What is Machine learning using an example?

For example, a robot that can work according to situations without giving outer instructions. and another example is that we all know the site amazon.com which is a very popular eCommerce site. so how Amazon recommended products to their customers? it’s based on machine learning. it uses the customer’s browsing and purchasing history and gives us product suggestions.

Types of Machine Learning

Based on the environment setup there are mainly three types of machine learning models that can be used to train an ML model. 

1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement Learning

Supervised Learning

In supervised learning, we train a model using some pre-trained data. first, we give some inputs which are pre-labeled and then train the model on the desired output. usually, we make rules and just map inputs to the output.

let’s take an example to understand it.

For example: let’s say we are training a model which predicts that whether there is a dog in a given image or not. then to train a model we first give some images as input. and we already know whether there is a dog or not in a particular image. and if the model gives us the wrong output then we train the model on the basis of the desired output. and correct the corresponding output. so the model will be able to take a picture and determine whether or not it contains a dog.  

Unsupervised Learning

In unsupervised learning, we don’t have any pre-labeled data. the model can find them on its own and find the structure in the given input and produce the output. here we use the search patterns.

For example: let’s take the same example. so now using unsupervised learning the model can predict whether there is a dog in pictures or not. eg differences in pixel color or orientations.

Reinforcement Learning

In reinforcement learning, we have a dynamic environment in which we perform different goals.

Example: playing a game against anyone. like a machine playing chess against a man.

There are other categories of machine learning which depend on the output.

Classification

In classification, we divided the inputs into different classes. for example, spam email filtering is a classification technique.

Regression

In Regression, we produce continuously valued outputs. for example, predict the house rent or electricity bill.

Clustering

In clustering, we divide the unlabeled inputs into different groups. for example, customer segmentation.

Density estimation

In DE we find the distribution of inputs in a particular space. like, predicting the result of class students.

Dimensionality reduction

In dimensionality reduction, we simply map the inputs in lower-dimensional space.

Note: if we get large enough data that is pre-labeled. then we use supervised learning. but it is not possible to get the fully labeled data to train a model. so here unsupervised or semi-supervised learning comes into use.

It is the first step we need to choose which type of learning we are going to use to train our model. whether it’s supervised, unsupervised or semi-supervised learning. 

What is semi-supervised learning?

When we use some labeled or unlabeled data set to train our model. then it comes to semi-supervised learning.

After choosing the right learning we also need to choose the proper model and most importantly be able to process data into a pipeline.