Machine Learning Types
Supervised Learning
model is trained on a labeled dataset (i.e., the target or outcome variable is known)
Examples:
- Image classification (cat vs dog)
- Spam detection
Unsupervised Learning
The model learns from unlabeled data (it finds patterns on its own)
Examples:
- Clustering (grouping similar items together)
- Anomaly detection (identifying unusual in data)
- Dimensionality reduction (reducing number of features(variables) while keeping the important information)
Self-Supervised Learning (SSL)
Models to train themselves on unlabeled data, creates its own labels automatically from the data itself.
Explanation:
- Instead of humans labeling data (like “this is a cat”), the model takes part of the data, hides it, and learns to predict the missing part.
Reinforcement Learning
Model learns by trial and error using rewards and penalties.
Examples:
- Robotics (robot learns to navigate environment)
- Game playing agents (Chess, Atari games)
Semi-Supervised Learning
combination between supervised and unsupervised learning.
Example:
Imagine you have:
- 500 labeled images (you know the correct category)
- 50,000 unlabeled images (no labels)
A semi-supervised model:
- Learns structure from all 50,000 images.
- Uses the 500 labeled examples to fine-tune the decision boundaries.
Result:
- You get performance close to a fully supervised model without labeling everything.
Why it matters:
- Labeling data is expensive and time-consuming. Semi-supervised learning leverages the alot of unlabeled data to improve model performance with minimal labeled data.
sources: