Types of Machine Learning

• 1. Supervised Learning
• 2. Unsupervised Learning
• 3. Semi-Supervised Learning
• 4. Reinforcement Learning

1. Supervised Learning (Labeled)

Supervised learning is a type of machine learning where the model is trained on a labeled dataset, meaning the input comes with the correct output. The goal is for the model to learn the mapping from inputs to outputs.

Example: Predicting house prices based on features like size, number of rooms, and location. The dataset includes both the input features and the actual prices (labels).

Use cases: Email spam detection, credit scoring, image classification.

Sample: Predicting house price

inputs = [[1000, 3], [1500, 4], [800, 2]]  # Size and number of rooms
labels = [200000, 300000, 150000]          # Actual prices
    

2. Unsupervised Learning (Unlabeled)

Unsupervised learning involves training on data without labels. The algorithm tries to identify patterns, structures, or groupings within the data.

Example: Customer segmentation: Grouping customers based on purchasing behavior without knowing their category in advance.

Use cases: Clustering, anomaly detection, market basket analysis.

Sample: Customer spending behavior

data = [[300, 2], [1000, 10], [250, 1], [1100, 11]]  # [Total spend, Number of visits]
# Output: Groups of similar customers
    

3. Semi-Supervised Learning (Mostly Unlabeled)

Semi-supervised learning is a mix of supervised and unsupervised learning. It uses a small amount of labeled data along with a large amount of unlabeled data to improve learning accuracy.

Example: Classifying web pages where only a few pages are labeled as “news” or “sports,” and the rest are unlabeled.

Use cases: Medical imaging, text classification when labeling is expensive.

Sample: Web page classification

labeled_data = [['text1', 'news'], ['text2', 'sports']]
unlabeled_data = ['text3', 'text4', 'text5']  # No labels
    

4. Reinforcement Learning (Inreraction with environment)

Reinforcement learning is based on agents learning to make decisions by performing actions and receiving feedback in the form of rewards or penalties.

Example: A robot learning to walk. It gets positive rewards for staying upright and moving forward, and negative rewards for falling.

Use cases: Game AI (like AlphaGo), robotics, dynamic pricing, self-driving cars.

Sample: Game playing

state = "game state"
action = "move right"
reward = +1 or -1  # Based on success/failure
    

Supervised vs. Unsupervised vs. Semi-Supervised vs. Reinforcement Learning

Feature	Supervised Learning	Unsupervised Learning	Semi-Supervised Learning	Reinforcement Learning
Data Type	Labeled	Unlabeled	Mostly Unlabeled	Interaction with environment
Goal	Predict outcomes	Find structure	Improve with fewer labels	Maximize reward
Feedback	Direct (ground truth)	No feedback	Limited feedback	Delayed reward
Examples	Classification, Regression	Clustering, Association	Image labeling, NLP	Robotics, Game AI