Supervised learning
지도학습
Learning from data where each input comes with the right answer.
If each photo comes with its label, that is supervised learning. Most classification and prediction lives here.
Someone has to provide those answers, which costs money. Where expert judgement is needed, as in learning material, labelling is where quality is decided.
If the definition of correct wavers, so does the model. Decide what counts as right before labelling begins.
- What it needsA labelling standard and a review process.
- The counterpartUnsupervised learning finds structure without answers.