Lesson 7-1 explained: The Basics of Machine Learning

Lesson summary

Machine learning finds the rules from data instead of being given them. It comes in three forms — supervised (learn from labeled examples), unsupervised (find hidden groups), and reinforcement (learn from rewards) — but its results depend on the data, so bias and the black-box problem make it a support for human judgment, not a replacement.

Takeaway: Machine learning finds rules from data: supervised (labeled → classification with features), unsupervised (unlabeled → clustering), reinforcement (rewards). Mind its limits — bias and the black-box problem — and keep it supporting human judgment.

Key terms

• Machine learning
• Bias
• Black-box problem
• Supervised learning
• Classification
• Feature
• Clustering
• Reinforcement learning
• Reward
• Label
• Human-in-the-loop

Textbook page: Arabic book 50 · English book 57

Video lesson

The AI black box and the transparency problem

Free questions on this lesson

From the textbook question bank. Think of the answer first, then open it.

1. 1. What is machine learning?

   • A program that runs rules written by a human step by step
   • A device that stores data without processing it
   • A way of drawing data in charts
   • A technology that learns patterns and regularities from large amounts of data to judge and predict on new data
   Answer

   Correct answer: A technology that learns patterns and regularities from large amounts of data to judge and predict on new data

   Machine learning finds patterns in lots of data, then uses them to judge and predict on data it has not seen.

   1. The Idea of Machine Learning · page 58

2. 2. What data does reinforcement learning learn from?

   • Actions and their results (rewards)
   • A ready list of rules
   • Inputs only, with no results
   • Pairs of inputs and labels
   Answer

   Correct answer: Actions and their results (rewards)

   Reinforcement learning tries actions, receives rewards, and learns from them how to choose better actions.

   2. Three Types of Learning Methods · page 59

3. 3. What is the key difference between classification and clustering?

   • Classification is only for numbers; clustering only for text
   • Classification makes few groups; clustering makes many
   • Classification learns from data with correct labels; clustering from data without labels
   • Classification is unsupervised; clustering is supervised
   Answer

   Correct answer: Classification learns from data with correct labels; clustering from data without labels

   The difference is one word: labels. Classification is given labels in advance; clustering is not.

   4. Classification vs clustering · page 61

Lessons in this chapter 7: Machine Learning and Artificial Intelligence

1. 7-1 The Basics of Machine Learning
2. 7-2 Neural Networks and Deep Learning
3. 7-3 Large Language Models (LLM) and Generative AI