3 Types of Machine Learning

Machine Learning is the go-to toolbox of the current business operations in a variety of domains. The implementation of machine learning into such operations is a strategic step and requires a lot of resources. Therefore, it is essential to understand what kind of business task you want to your Machine Learning algorithm to work upon. You must also understand the different types of perks and flavors they bring to the table.

Based on the different flavors and objectives that a business can have, these machine learning algorithms are broadly classified as:

  • Supervised Learning – “Teach me what to learn”
  • Unsupervised Learning – “I will find what to learn”
  • Reinforcement Learning – “I’ll learn from my mistakes at every step (Hit & Trial!)”

What is Supervised Learning?

The supervised learning approach is similar to human learning under the supervision of a teacher. The teacher provides good examples for the student to memorize (learn), and the student then derives general rules from these specific examples to use on a new example.

In other words, this algorithm learns from example data (training data) and associated response (target). It is done to predict the correct response when given a new example (test data).

Some might also call this a “spoonfed” approach of machine learning. You select the kind and extent of data that needs to be inputted (fed) to the algorithm. You also decide what type of answer you desire (for example, True/False or Yes/No).

Real-life examples:

  • Email Spam (Classification)– The algorithm takes historical spam and non-spam emails as input. Consequently, it draws patterns in data to classify spam from others.
  • Stock Price Prediction (Regression)– Historical business market data is fed to the algorithm in this method. With proper regression analysis, the new price for the future is predicted.

The most widely used supervised learning approaches include:

  • Linear Regression
  • Logistic Regression
  • Decision Trees
  • Gradient Boosted Trees
  • Random Forest
  • Support Vector Machines
  • K-Nearest Neighbors etc.

What is Unsupervised Learning?

Sometimes a kid does not need explicit supervision of a teacher. Instead, she tries to learn on its own based on her past experiences with the world. She memorizes (learns) what she has observed with the previous lessons (training data) and tries to replicate that in an unseen scenario (test data).

In simple words, this learning occurs when an algorithm learns from plain examples without any associated target (response). It leaves the algorithm to determine the data patterns on its own.

For Example:

Data with similar traits are asked to group by the algorithm. These groups are called clusters, and the process is called clustering. In retail analytics, various customers are usually clustered based on their purchase and other behaviors.

Popular techniques used in Unsupervised Learning:

  • k-means clustering
  • t-SNE (t-distributed Stochastic Neighbor Embedding)
  • PCA (Principal Component Analysis)
  • Association Rule

What is Reinforcement Learning?

This form of learning is totally different from what we have discussed above. Sometimes, a kid might neither have any experiences nor any supervision to learn. But imagine that she has a godfather who gives her a candy whenever she learns something useful and punishes her when she makes a mistake. Alternatively, her training is now aided by rewards (candy) and punishments (corrections). This process makes her learn from her mistakes about where and how to improve. This is called as Reinforcement Learning. In this method, you reinforce the learning curve of the model.

It is a form of reward/punishment-based learning in which a machine learns a series of actions to perform a task. Each action of this algorithm is tied with a positive (candy) or negative (punishment) reward.

An exciting example of reinforcement learning occurs when computers learn to play video games by themselves. The algorithm keeps on interacting with the game environment through a series of actions. This environment, in turn, gives a reward or punishment based on the nature of action taken.

Most common reinforcement learning algorithms include:

  • Q-Learning
  • Temporal Difference (TD)
  • Monte-Carlo Tree Search (MCTS)
  • Asynchronous Actor-Critic Agents (A3C)


I hope by now, you have understood the kinds of machine learning methods that we have at our disposal. We saw that there is a multitude of techniques developed in this domain, each to tackle a specific flavor of the objective. Further, we dealt with three major categories in which these can be classified.