It earns a place when
- Use the setting to choose data collection, evaluation and model families.
- You can evaluate it against a credible baseline.
- Its output fits the decision and data constraints.
Machine learning models · Phase 1 · Lesson 2
Learning settings and sources of supervision
Classify supervised, unsupervised, self-supervised and reinforcement learning by the signal that drives learning.
Lesson goal
Classify supervised, unsupervised, self-supervised and reinforcement learning by the signal that drives learning.
The core idea
Ask where the target or feedback comes from. A human label, structure in the data, a target constructed from the data, and a reward are different training situations.
Mental model
Picture it this way. Ask where the target or feedback comes from. A human label, structure in the data, a target constructed from the data, and a reward are different training situations. The important question is what assumption this picture makes, and whether that assumption fits the data.
Mathematical core
Supervised learning minimises a loss over pairs (x, y). Reinforcement learning instead estimates value over state, action and future reward.
Worked example
Grouping customers without labels is unsupervised. Masking words and predicting them is self-supervised. Learning a game policy from score changes is reinforcement learning.
When to use it
Do not make it the default when
- Do not infer the setting from whether a neural network is used. Neural networks can appear in all four settings.
- A simpler model has not been tested.
- The data or target definition is still unclear.
Failure modes
Watch for this. Self-supervised targets can be easy to optimise but misaligned with the downstream task. Rewards can be delayed or sparse.
When a result looks surprisingly good, inspect the split, target timing, error slices and data-generating process before celebrating.
Practice
Use this as a small experiment rather than a recipe to copy blindly. Change one thing, record the result and explain the change.
Do this. Take four tasks from your work or life and classify each by its training signal. For each, state what the target or feedback would be.
Retrieval check
Answer from memory first. The buttons reveal feedback, but the durable step is explaining why.
1. What defines supervised learning?
2. Why is masked-word prediction called self-supervised?
3. What is distinctive about reinforcement learning?
Transfer prompt. Describe one real problem where this model or idea would be a sensible candidate. Name the target, the main risk and the metric you would inspect.
Primary source
Google ML Crash Course. Use the source for the deeper treatment after you can explain the lesson's core idea without looking.