Supervised Learning and Unsupervised Learning - Course on Deep Learning

Welcome back Readers!! Let us begin today’s article without wasting much time. Today we will learn about various Machine Learning algorithms.

Types of Machine Learning Algorithms

As seen in Fig. 1, various machine learning algorithms exist in literature. So, let us simplify the discussion by learning them according to levels. Level 1 comprises supervised learning and unsupervised learning. Level 2 comprises classification-based techniques and regression-based techniques which uses supervised learning, & clustering (hard and soft) techniques and dimensionality reduction techniques which uses unsupervised learning. Finally, we will study their individual types in detail in upcoming articles. But this article is not finished yet, keep reading.

Fig. 1 Classification of Machine Learning Algorithms


  • Supervised Learning: 

Just image that you are preparing for Mathematics exam. You are given a definite syllabus and told that if you will study this syllabus, you will be able to answer the questions in examination. So, what strategy will be followed by not so bright students (like me) a couple of days before exam? Well, I will just have a look at formulas and try to solve a problem or two on each formula. And when doomsday will arrive (Sorry!! I meant examination day 😅), the questions will similar to those which we practiced, only the numeric values will be different. So, I will apply similar formula to those questions and get the answers. This is exactly what is happening in supervised learning. In supervised learning, a model is trained using inputs called training data and outputs called responses. We teach machine that if we are giving some input (say X), we are getting some output (say Y). And once the machine is trained, we use test data to check if the prediction made is correct/desirable or not.

  • Unsupervised Learning: 

Now imagine a scenario, you entered a new class or new semester of your school/college. And on the first day itself, teacher enters the classroom and says "Students, today we are going to have a surprise test." What thoughts will cross your mind: "We know nothing", "How are we gonna pass this subject", "We don't even know the syllabus", etc. Right!! This is exactly how a machine feels if it is unsupervised. No training data is present, no goal is fixed, the dimensions of data is reduced. Why do we even use this type of learning? We will get to the answer, but have to move step by step.


  • Classification:

We all have attempted true/false questions and multiple choice questions at some point in our life. Classification is no different than that. The outputs are placed in classes and labels based on prediction.  The problems can be of binary classification (eg. True/False or Yes/No), multi-class classification, and multi-label classification.

  • Regression:

When the output is predicted using the input, then we can say that we are working on regression problems. For instance, you have an image captured in rain. Using regression based technique, you can de-rain the image.

  • Clustering:

Cluster means a group, which can be of same or different elements. When each data belongs to only one group, then we deal with hard clustering problem. But, when some data belongs to two or more groups, then we deal with soft clustering.

  • Dimensionality Reduction:

As the name implies, there is reduction in dimensions of the input data. Dimensions are formed with features. More the number of features, more is the dimension of data. For larger dataset, more features are extracted which is not feasible. But what actually is this FEATURE and how it is EXTRACTED?

Stay tuned to learn more in detail. We will come back with much more information and interesting facts.

QUESTION ASKED IN INTERVIEWS: What is the difference between class and label in machine learning?

ANSWER: We have three items, namely cats, dogs, and apples which are categorized under animals and fruits. Cats, Dogs, and Apples are labels. Animals and Fruits are classes.

Hope you like this article. You can give your feedback about this article of Deep Learning Series and suggestions for the future articles in the below mentioned comment box 👇. Thank you for your cooperation throughout the year 2021. This will be the last article of this year. The series will continue in January, 2022. Keep supporting us like this, it will motivate us to give more than our 100%.


  1. The work done by you is very good.
    Get to learn a lot. Good job 👌👍

    1. Dear Sir/Ma'am,
      Glad you like the article. Keep supporting us !!

  2. Replies
    1. Dear Sir/Ma'am,
      Glad you like the article. Keep supporting us !!


Powered by Blogger.