Types Of Machine Learning
⭐ Types of Learning in Machine Learning ⭐
Machine Learning systems learn from data. Depending on how the data is used to train the model, ML is of different types. This chapter covers all 4 types — Supervised, Unsupervised, Semi-Supervised & Reinforcement Learning with examples, algorithms, use cases & comparison. On Data Insights.
📑 Topics Covered:
- Supervised Learning
- Unsupervised Learning
- Semi-Supervised Learning
- Reinforcement Learning
- Comparison of Different Types
- When to Use Which Type?
What are Types of Learning?
Based on the nature of data and the way a model learns, Machine Learning is mainly divided into four types:
1. Supervised Learning
The model is trained on labeled data. It learns the mapping from input to output.
2. Unsupervised Learning
The model is trained on unlabeled data. It finds patterns or structure in data.
3. Semi-Supervised Learning
The model is trained on both labeled and unlabeled data.
4. Reinforcement Learning
The model learns by interacting with an environment and getting rewards.
Data + Experience → Learning → Patterns → Better Predictions & Decisions
Why Different Types of Learning?
- Different problems require different learning approaches.
- Availability of data (labeled or unlabeled) decides the type.
- Each type has its own use cases, strengths and limitations.
- Choosing the right type improves performance and accuracy.
Learning Types Overview
Learning
Learning
Learning
Learning
1. Supervised Learning
In supervised learning, the model is trained using labeled data. Each training example has input features and the correct output.
🔧 How it works?
| Input (Features) | Algorithm (Learning) | Output (Target/Label) | Prediction (New Data) |
|---|---|---|---|
| X₁, X₂, X₃, ... | → Learns pattern → | Y | Ŷ (predicted) |
📊 Example:
Predict employee salary based on features:
| Experience (years) | Department | Performance | Salary (₹) |
|---|---|---|---|
| 3 | IT | Good | 55,000 |
| 5 | HR | Excellent | 72,000 |
| 4 | Finance | Average | 65,000 |
| 6 | Marketing | Excellent | 80,000 |
🤖 Common Algorithms:
🎯 Use Cases:
2. Unsupervised Learning
In unsupervised learning, the model is trained using unlabeled data. The model tries to find hidden patterns, groups or structure.
🔧 How it works?
| Input Data (Unlabeled) | Algorithm (Finding Patterns) | Patterns/Structure (Clusters, Groups) |
|---|---|---|
| Raw data without labels | → Discovers structure → | Groups, Clusters, Rules |
📊 Example:
Employee segmentation in a company:
| Name | Experience | Salary (₹) | Performance Score |
|---|---|---|---|
| Aarav | 3 | 55,000 | 78 |
| Ishita | 5 | 72,000 | 92 |
| Kabir | 4 | 65,000 | 85 |
| Diya | 2 | 58,000 | 70 |
| Rohan | 6 | 80,000 | 95 |
↑ No "label" column — model finds groups on its own (e.g., high performers vs average)
🤖 Common Algorithms:
🎯 Use Cases:
3. Semi-Supervised Learning
In semi-supervised learning, the model is trained using a small amount of labeled data and a large amount of unlabeled data.
🔧 How it works?
| Labeled Data | Algorithm (Learning) | Unlabeled Data | Model (Prediction) |
|---|---|---|---|
| Small set (labeled) | → Learns from both → | Large set (unlabeled) | Better predictions |
📊 Example:
Employee performance prediction:
| Name | Experience | Salary (₹) | Performance Label |
|---|---|---|---|
| Aarav | 3 | 55,000 | Good ✅ (labeled) |
| Ishita | 5 | 72,000 | Excellent ✅ (labeled) |
| Kabir | 4 | 65,000 | ❓ (unlabeled) |
| Diya | 2 | 58,000 | ❓ (unlabeled) |
| Rohan | 6 | 80,000 | ❓ (unlabeled) |
↑ Only 2 employees have labels, rest 3 are unlabeled. Model learns from both!
🎯 Use Cases:
4. Reinforcement Learning
In reinforcement learning, the model (agent) learns by interacting with an environment. It gets rewards for good actions and penalties for bad actions.
🔧 How it works?
| Agent | State | Action | Environment | Reward (World) |
|---|---|---|---|---|
| Model | Current position | Decision made | Real world | +1 / -1 |
📊 Example:
Employee task assignment optimization:
| Agent Action | Environment Response | Reward |
|---|---|---|
| Assign Aarav to Python project | Aarav has Python skills → Completed on time | +1 ✅ (Good match) |
| Assign Ishita to coding task | Ishita is HR, not developer → Delayed | -1 ❌ (Bad match) |
| Assign Kabir to finance report | Kabir is Finance → Delivered perfectly | +1 ✅ (Perfect match) |
↑ Agent learns from rewards — next time better task assignments!
🎯 Use Cases:
Comparison of Different Types
| Type | Data Required | Label Needed? | Goal |
|---|---|---|---|
| Supervised Learning | Labeled | Yes | Predict output for new data |
| Unsupervised Learning | Unlabeled | No | Find patterns or structure |
| Semi-Supervised | Both | Few labels | Learn from limited labels |
| Reinforcement | Interaction | No (rewards) | Improve model by trial & error |
When to Use Which Type?
| Scenario | Best Type | Why |
|---|---|---|
| Predict employee salary | Supervised | Have labeled salary data |
| Group similar employees | Unsupervised | No labels, find patterns |
| Some employees rated, most not | Semi-Supervised | Mix of labeled + unlabeled |
| Optimize task assignments | Reinforcement | Learn from outcomes |
| Spam email detection | Supervised | Labeled spam/not-spam |
| Customer segmentation | Unsupervised | Find natural groups |
| Game playing AI | Reinforcement | Trial and error learning |
| Medical image + few labels | Semi-Supervised | Expensive to label all |
""
MADE WITH 100% DEDICATION BY → DATA INSIGHTS
Next: ML Handbook — Chapter 3
Next chapter mein hum cover karenge: Supervised Learning Deep Dive — Regression vs Classification, Linear Regression, Logistic Regression, Decision Trees, aur real-world implementation. Data Insights par!
Happy Learning & Keep Exploring! 🚀
💬 Comments (0)
Loading comments...