Types of Machine Learning

⏱️ 15 min 📚 Lesson 2 of 11
1 / 10
👋

Welcome Back!

After exploring how computers can learn patterns from data to make predictions and decisions, we now examine the main ways ML approaches learning. You’ll delve into supervised, unsupervised, and reinforcement learning, seeing how each method handles data differently to solve various problems.

📊

Machine Learning Categories

There are several ML approaches: Supervised, Unsupervised, and Reinforcement learning. Each handles data differently.

In supervised learning, data comes with correct answers (labels) so the model learns by example. In unsupervised learning, data has no labels, and the model finds hidden patterns on its own.

Reinforcement learning involves an agent learning by trial and reward.

👨‍🏫

Supervised Learning

In supervised learning, every training example has a label: the correct answer. It's like having a teacher: for each input, we know the output we want.

For example, we might train on labeled photos where each image is tagged 'cat' or 'dog'. The model learns to connect features to those labels.

Once trained, it can predict labels for new, unseen examples.

🔍

Unsupervised Learning

Unsupervised learning works with data that has no labels. The model explores the data and finds patterns by itself.

Imagine sorting a pile of mixed mail into groups by similarity without knowing the destinations. The algorithm might group customers by buying habits or group articles by topic.

It's discovering the structure in the data.

🎯

Clustering (Unsupervised)

Clustering is a common unsupervised method. It groups similar items together.

For example, given points on a scatterplot, the algorithm will identify clusters (like circles around dense regions). Think of grouping fruits by color and shape without knowing their names.

Clustering helps us organize data into meaningful groups without predefined labels.

🎮

Reinforcement Learning

Reinforcement learning is like trial-and-error learning. An agent (computer) makes decisions and gets a reward for good actions.

It's like training a pet: give it a treat when it does a trick correctly. Over time, the agent learns which actions lead to the highest total rewards.

For example, an AI can learn to play a video game by receiving points (rewards) for good moves and negative feedback for mistakes.

⚖️

Comparing the Types

To recap: Supervised = learn with a teacher (labels provided); Unsupervised = find hidden patterns without labels; Reinforcement = learn by rewards and penalties.

Each type suits different tasks. We will see supervised and unsupervised tasks in more detail soon.

Quiz: Learning Types

Which Scenario Is An Example Of Unsupervised Learning?

A
A model trained on labeled cat/dog images
B
A robot learning to play a game through rewards
C
Grouping news articles by topic without labels
D
Training a spam filter with labeled emails

Fill in the Blank

In supervised learning, the provided correct answers are called ___.

💡 Drag the correct word from below into the blank to complete the sentence.
In supervised learning, the provided correct answers are called
Features
Clusters
Labels
Rewards

Reflection

💭

Think of a problem where you have many examples but no obvious answers. How might unsupervised learning help you?

Conversely, imagine a task like learning a game. How could reinforcement learning (trial and reward) apply?

Describe each scenario.

Lesson Completed!

Excellent work! You now understand the three main types of machine learning. Keep going to dive deeper!