Part 3: Types of Machine Learning

Think about how you learned things growing up-being taught by someone, figuring things out on your own, or learning from mistakes.

Machine Learning follows these same patterns

In this blog, we'll break down the types of Machine Learning and explore how machines learn from data

⚙️Types of Machine Learning

Machine Learning can be divided into different types based on how models learn from data.

Let's explore each of these types in detail.

1. Supervised Learning

Supervised Learning is a type of Machine Learning where the model learns from labeled data, i.e., the correct output is already given.

How it works

Imagine a student is learning from a teacher. The teacher gives the questions along with its answers and points out mistakes. Over time, the student starts recognizing patterns and improves gradually.

In Supervised Learning the model learns the same way.

Example

Predicting the house prices based on features like location, condition, and number of rooms available.

Here the model is trained on labeled features and then tested on new unseen data.

Types of problems

1. Regression (predicting continuous values, e.g., house price prediction)

2. Classification (predicting categories, e.g., spam and not spam emails)

Common Algorithms

1. Linear Regression

2. Logistic Regression

3. KNN

4. Decision Trees

5. Support Vector Machines (SVM)

Use Cases

1. Email Spam Detection

2. Stock price prediction

3. Medical diagnosis

2. Unsupervised Learning

Unsupervised Learning is a technique where models are trained on unlabeled data. It is used to find hidden patterns and relationships by grouping similar data together.

How it works

Imagine you walk into a party where you don't know anyone. As you observe, you start noticing groups forming-people with similar interests, professions, or personalities naturally gather together.

No one tells you who belongs where, but patterns still emerge.

Unsupervised Learning works the same way-it identifies hidden patterns and groups in data without any predefined labels.

Example

Grouping customers based on their shopping behavior.

Types of tasks

1. Clustering (Grouping similar data points)

2. Dimensionality Reduction (simplifying data while keeping important information)

Common Algorithms

1. K-Means Clustering

2. Hierarchical Clustering

3. Principal Component Analysis (PCA)

Use Cases

1. Customer Segmentation

2. Anomaly Detection (fraud detection)

3. Market Research

3. Semi-Supervised Learning

Semi-Supervised Learning is a combination of supervised and unsupervised, where a small amount of labeled data and large amount of unlabeled data is used for model training.

Here the model first learns from labeled data and then improve using unlabeled data.

How it works

Imagine you are shown how to solve a particular set of problems, and than are given many similar ones to solve.

You start to recognize and learn patterns by solving the problems and then complete the rest accordingly.

Example

Image classification where few images are labeled, but thousands are not.

Use Cases

1. Medical Imaging

2. Speech Recognition

3. Large-scale classification problems

4. Reinforcement Learning

Reinforcement Learning is a technique where an agent learns by interacting with an environment and receiving rewards and penalties.

Here the model learns through trial and error method, aiming to maximize rewards over time.

How it works

Think about the first time you rode a bicycle. There were no specific instructions for every movement-you try, fall, adjust, and improve with practice.

Example

Training an AI to play a game of chess. For every correct move points are given and for every incorrect move points are deducted.

Common Algorithms

1. Q-Learning

2. Deep Q Learning (DQN)

Use Cases

1. Game AI

2. Robotics

3. Self-driving cars

🚀Final Thoughts

Each type of Machine Learning represents a different way of learning-just like humans. Whether it's learning from examples, exploring patterns, or improving through trial and error, these approaches shape how machines make decisions.

Understanding these different learning approaches helps build a strong foundation for seeing how machines learn and make decisions.

🔎Coming Up Next

In the next blog, we'll dive deeper into Supervised Learning and understand how Regression and Classification helps in solving real-world problems.