Back to library

🏷️Supervised vs Unsupervised Learning

Stop memorizing the labels-vs-no-labels split. Learn to classify any ML problem by where its supervision comes from — including the messy self-supervised middle that powers modern AI.

Foundations14 drops~2-week path · 5–8 min/daytechnology

Phase 1What 'Labels' Actually Mean

See why labels create the supervised/unsupervised divide

4 drops
  1. Labels are answer keys, not data

    6 min

    Labels are answer keys, not data

  2. Spam filters: the cleanest supervised case

    6 min

    Spam filters: the cleanest supervised case

  3. Customer segmentation: discovering groups you didn't define

    6 min

    Customer segmentation: discovering groups you didn't define

  4. The two-regime split is a lie we tell beginners

    7 min

    The two-regime split is a lie we tell beginners

Phase 2Classifying Real ML Problems

Classify ten real ML problems including borderline cases

5 drops
  1. Five problems with obvious answers

    6 min

    Five problems with obvious answers

  2. Recommendation systems break the rules

    7 min

    Recommendation systems break the rules

  3. Anomaly detection: regime depends on history

    7 min

    Anomaly detection: regime depends on history

  4. Search ranking is supervised — by clicks

    7 min

    Search ranking is supervised — by clicks

  5. Reinforcement learning isn't either

    7 min

    Reinforcement learning isn't either

Phase 3Beyond the Binary: Self-Supervised and In Between

Decode self-supervised, semi-supervised, and pretext labels

4 drops
  1. Self-supervised: the data labels itself

    7 min

    Self-supervised: the data labels itself

  2. Pretext tasks: the trick that broke the binary

    7 min

    Pretext tasks: the trick that broke the binary

  3. Semi-supervised: when labels are precious

    7 min

    Semi-supervised: when labels are precious

  4. The full taxonomy: where every technique fits

    7 min

    The full taxonomy: where every technique fits

Phase 4Designing Both Regimes for Your Problem

Design supervised and self-supervised setups for one problem

1 drop
  1. Design supervised and self-supervised for one real problem

    8 min

    Design supervised and self-supervised for one real problem

Frequently asked questions

What's the actual difference between supervised and unsupervised learning?
This is covered in the “Supervised vs Unsupervised Learning” learning path. Start with daily 5-minute micro-lessons that build from fundamentals to hands-on application.
Is self-supervised learning supervised or unsupervised?
This is covered in the “Supervised vs Unsupervised Learning” learning path. Start with daily 5-minute micro-lessons that build from fundamentals to hands-on application.
Why does ChatGPT count as self-supervised if it predicts the next word?
This is covered in the “Supervised vs Unsupervised Learning” learning path. Start with daily 5-minute micro-lessons that build from fundamentals to hands-on application.
When should I use supervised learning over unsupervised?
This is covered in the “Supervised vs Unsupervised Learning” learning path. Start with daily 5-minute micro-lessons that build from fundamentals to hands-on application.
How do you handle a problem with only a few labels and lots of unlabeled data?
This is covered in the “Supervised vs Unsupervised Learning” learning path. Start with daily 5-minute micro-lessons that build from fundamentals to hands-on application.