Types of Machine Learning
Explore the three primary machine learning paradigms: supervised learning with labeled data for prediction, unsupervised learning for discovering patterns without labels, and reinforcement learning where an agent learns through interactions and rewards. Understand how to choose the right approach based on data and problem type to apply machine learning effectively.
Machine learning (ML) is an expansive field, but at its heart we are teaching computers to learn from data without being explicitly programmed.
Not all problems look the same, and the process we use to build a spam filter, which needs labeled examples of “spam” versus “not spam,” is fundamentally different from the process we use to organize millions of customer photos, which have no predefined labels. To successfully apply predictive modeling, we first need to answer a critical question: What type of data do we have, and what are we trying to achieve?
In this lesson, we will introduce the three main learning paradigms, supervised, unsupervised, and reinforcement learning, so we can choose the correct approach for any given problem.
The role of data in machine learning
Data is the essential fuel that allows machine learning models to grasp
For example, to build a facial recognition system, the required data consists of numerous face images and their corresponding identities (known as labels). The ultimate goal is to learn a mapping function that accurately associates each face image with its correct identity label. This process, where the output is known and provided to the model, is said to have
Based on the type of data available and the nature of the learning problem, machine learning can be broadly categorized into three main types:
- Supervised learning
- Unsupervised learning
- Reinforcement learning
Supervised learning
While all machine learning algorithms utilize data, supervised learning is considered the most data-intensive category.
A supervised learning algorithm takes labeled data as input and uses it to train a model that can make predictions or inferences on new, unlabeled data. A model is a mathematical representation of the relationship between the input features and the target class. The input features (independent) refer to the quantifiable attributes that describe the data, such as weight, height, and color, while the target class (dependent) is the desired output, such as identifying spam or non-spam emails. Hence, the main job of this type of algorithm is to find the relation between input features and their respective target class so that the trained model can predict the label of the unseen data in the future.
In the following animation, the model first learns by being trained on labeled data (shapes associated with numeric labels 1, 2, 3). Once trained, the machine can accurately apply the learned relationship to classify or predict the labels for new, unseen unlabeled data (the shapes without initial numbers).
Common supervised learning applications
There are two main types of supervised machine learning: regression and classification.
-
Regression is the type of supervised learning where the goal is to predict a continuous output variable. For example, when predicting the price of a house, the input might be the features of a house, such as the number of bedrooms and bathrooms, and the output will be the price of the house.
-
Classification is a type of supervised learning that aims to predict a categorical label for each input. For example, in fruit classification, the input is a fruit (this could be an actual image of the fruit or a list of its properties like weight, color, and shape), and the output is a label indicating whether the fruit is an apple or a mango.
Unsupervised learning
Unsupervised learning deals with grouping the data based on some similarity/dissimilarity. Unlike supervised learning, it extracts patterns from the given data without labels. In other words, this technique distinguishes possible clusters in the data based on some similarity matrices. As labels aren’t required for this method, it requires minimal human intervention.
In the following animation, the model takes in unlabeled scattered data (mixed dots) and, without prior examples or guidance, identifies hidden similarities within the data. It then groups the inputs into defined categories or clusters (Cluster 1, Cluster 2, Cluster 3), autonomously revealing underlying structure and organization.
Common unsupervised learning applications
There are two main types of unsupervised machine learning tasks: clustering and dimensionality reduction.
Clustering groups similar data points together to discover distinct segments (like customer groups that buy frequently and spend highly or infrequently with low average purchases).
Dimensionality Reduction simplifies data by reducing the number of input features and removing irrelevant features (like using only a student's final exam score and total attendance instead of their scores on every quiz, homework, and class participation if the latter aren't influencing the final results).
In cases where we have a mix of labeled and unlabeled data, the training process is referred to as semi-supervised learning. It uses the small labeled portion to guide learning while still leveraging the large volume of unlabeled examples.
Reinforcement learning
In reinforcement learning, an agent (the learner) learns to take actions in an environment to maximize a reward signal. It’s a unique type of learning, characterized by continuous interaction. For example, a self-driving car is learning to navigate city streets. It decides when to accelerate, brake, or turn, while the environment, including traffic, pedestrians, and road conditions, responds. The car receives rewards for staying safe, following traffic rules, and reaching its destination efficiently. Over time, it learns which actions lead to the best outcomes.
As depicted in the diagram below, the agent takes an action () within its environment. In response, the environment transitions to a new state () and provides a reward () to the agent. This cyclic interaction drives the learning process.
This feedback mechanism distinguishes RL. It’s not supervised, because there are no ‘correct’ labels for individual decisions. But it’s not unsupervised, because the agent does receive feedback in the form of rewards—they are just often delayed. Consider chess. An agent might make a move at the beginning of the game. It won’t know if that move was “good” or “bad” until 50 moves later, when it either wins or loses. The reward (winning) is delayed, and the agent’s challenge is to figure out which of its early actions were responsible for the final victory.
Common reinforcement learning applications
-
Self-driving cars: Learning to navigate traffic, pedestrians, and road conditions.
-
Game AI: Training agents to master games like Chess or Go.
-
Robotics: Teaching robots to perform manual tasks through feedback loops.
Comparsion
Machine learning paradigms each serve unique purposes depending on data type and business objectives. Understanding the differences helps select the most effective ML approach, optimize model design, and achieve actionable insights across domains such as finance, healthcare, e-commerce, and AI-driven automation.
Paradigm | Data Requirement | Goal | When to Use | Applications |
Supervised learning | Labeled data | Predict outcomes | Historical labeled datasets; regression/classification tasks | Spam detection, churn prediction, price forecasting, disease diagnosis |
Unsupervised learning | Unlabeled data | Discover patterns | When labels are unavailable, clustering/dimensionality reduction | Customer segmentation, anomaly detection, market analysis, feature compression |
Reinforcement learning | Feedback via rewards | Optimize actions over time | Sequential decision-making; delayed feedback | Self-driving cars, robotics, game AI, resource management |
Conclusion
We have explored the three foundational categories of machine learning, each defined by the nature of the data it requires and the goal it seeks to achieve.
Supervised learning uses labeled data to predict known outcomes, such as prices or categories. Unsupervised learning uses unlabeled data to discover hidden structures and groupings, such as customer segments. Reinforcement learning operates through trial and error, taking actions within an environment to maximize a delayed reward, as seen in applications such as game playing or robotics.
Understanding these distinctions is a critical first step toward proficiency in machine learning, as it enables us to frame real-world problems within the appropriate learning paradigm.
Note: While Reinforcement Learning is a foundational paradigm, this course will focus primarily on the deep dives into supervised and unsupervised techniques.