Supervised learning trains on examples that carry the answer, then predicts that answer for new cases.
What the base framework establishes
The public contents places supervised learning first among the approaches, which reflects its role as the reference case in the framework. The model is given inputs and the corresponding target, and learns a mapping that can be applied to new inputs. The quality of the labels sets the ceiling on what can be learned, and the split between training, validation and test data sets the honesty of the result. The framework's data and pipeline clauses carry the consequences: labels are a data preparation activity, and the test set must stay out of the training and tuning loop. The clause text is not in the public sample and is not restated here.
Why it matters to an AI management system
Supervised systems are the easiest to evidence and the easiest to overstate. A governance review should ask where the labels came from and whether the test set was ever seen during development. Those two questions expose most of the false confidence in an accuracy figure.
The public draft summary names no change to 7.2 Supervised machine learning. The final text could still touch it.
Evidence a reviewer samples
| Category | Artefact |
|---|---|
| Label provenance | Who or what produced each label and under which instruction. |
| Split record | The training, validation and test split, with dates and overlap checks. |
| Test-set integrity note | Confirmation that test data was not used for tuning. |
One question to prepare for. Who produced the labels, and has the test set ever influenced a design choice?
Common gaps
- Labels are assumed to be ground truth because they came from a business system.
- The test set was used repeatedly to choose the model, so the reported result is optimistic.
- Label quality is never measured, only the model's agreement with the labels.
Prepare before publication
Record label provenance and the split, and add a test-set integrity line to the evaluation report. Measure label agreement on a sample.
Dates that matter
No separate date attaches to 7.2; the record dates are the stage date, the planning estimate and the CEN expectation stated above.
Cross-references
3.3 Data · 6.2 Task · 8.3 Data preparation · 8.5 Verification and validation
Source
Source for 7.2: ISO/IEC 23053:2022, clause 7.2. The public sample stops before 7.2, so this page relies on the public contents and the public summary and does not quote the clause. The two official records are linked on this page so any line can be checked.
Not legal advice; does not certify. A reading of the framework and the evidence a reviewer commonly samples, not a substitute for the standard or your auditor's rules.