Volume 3: The Modern Frontier: Scale, Intelligence, and What Comes Next3/10
  1. 01Systems at Scale 201-210
  2. 02Modern Data Systems 211-220
  3. 03Machine Learning 221-230
  4. 04Generative AI Systems 231-240
  5. 05Security and Privacy 241-250
  6. 06Emerging Computing 251-260
  7. 07Technology and Society 261-270
  8. 08Digital Economics 271-280
  9. 09Responsible Futures 281-290
  10. 10Synthesis 291-300

Chapter 03 of 10

Machine Learning 221-230

10 lessons52 min readingParts 221-230
  1. 01Supervised LearningHow does a model learn from examples with known answers?5 min read
  2. 02Unsupervised LearningWhat can a model learn when examples have no supplied labels?5 min read
  3. 03Reinforcement LearningHow can an agent learn through actions and delayed rewards?6 min read
  4. 04Features and LabelsWhat information goes into a predictive model, and what answer should it learn?5 min read
  5. 05Training, Validation, and Test SetsWhy split machine-learning data into separate groups?5 min read
  6. 06Overfitting and UnderfittingWhy can a model perform well in training but poorly in the real world?5 min read
  7. 07Precision and RecallHow should a classifier be judged when different mistakes have different costs?5 min read
  8. 08Bias in Machine LearningHow can a model reproduce unfair or distorted outcomes?5 min read
  9. 09Model DriftWhy can a model become less useful after deployment even when its code never changes?5 min read
  10. 10MLOpsWhat operational work surrounds a production machine-learning model?6 min read