From Rules to Learning
Traditional programs follow fixed rules: if this, then that.
Machine learning (ML) is different:
- We give the computer data and examples
- It learns patterns by itself
- Then it makes predictions or decisions
> ML helps computers improve with experience instead of only obeying hand-made rules.
Everyday uses: recommendations, spam filters, voice assistants.
Key Idea: Learn From Data
In ML we talk about:
- Inputs: information we give (features)
- Outputs: what we want to predict (labels)
- Model: math function connecting inputs to outputs
Learning = adjusting the model so its predictions match real data as closely as possible.
Major Types of ML
1. Supervised learning
- Data comes with correct answers
- Example: email + label “spam / not spam”
2. Unsupervised learning
- No labels, just data
- Model finds structure or groups
3. Reinforcement learning
- Agent takes actions
- Gets rewards or penalties, learns by trial and error.
💡 This is just Chapter 1. The full content with all chapters, interactive quizzes, and progress tracking is available in the Octo AI app.