Training, validation and testing data

By Jorge García Aibar Updated 2026-09-25

The three data sets the AI Act defines for high-risk systems: training data fits the model, validation data tunes it and guards against overfitting, and testing data gives an independent check before release ¹. Keeping them separate, and proving it, is what stops test-set contamination.

Developed in: ch. 14, Data for training and testing
Chapters: ch. 14, Development
Contrast with: Test-set contamination
Source: 1 numbered reference, listed below

Where it is used

13 chapters of the Body of Knowledge use the term. Each link opens the first section that does.

01 · Definition The object of governance 1 mention
04 · The Stack Data governance across the stack 1 mention
08 · Regulatory Map EU AI Act, post-Omnibus 3 mentions
10 · Reading List Canonical papers: documentation, audit and accountability 3 mentions
13 · Risk Management Identifying risk: sources, factors and stakeholders 2 mentions
14 · Development The technical file 2 mentions
15 · Deployment Choosing the model 6 mentions
16 · Fairness & XAI Mitigation before, during and after training 3 mentions
17 · Incidents Root-cause analysis 2 mentions
18 · EU AI Act High-risk requirements (Articles 8 to 15) 2 mentions
19 · Privacy & AI Principles applied to AI 3 mentions
21 · AI Laws Worldwide South Korea: the AI Basic Act 4 mentions
22 · Principles & Standards Council of Europe Framework Convention (CETS No. 225) 1 mention

Patterns that use this term

4 pattern pages use the term, most mentions first.

AI Threat Model 5 mentions
Dataset Admission Gate 3 mentions
Downstream Use Register 2 mentions
Vendor / Model Due-Diligence Gate 1 mention

Sources

[1] Regulation (EU) 2024/1689 (AI Act), consolidated text as amended by Regulation (EU) 2026/1744 (Digital Omnibus on AI, in force 27 Jul 2026; definitions in Art. 3, incl. 3(1), 3(3) to 3(14), 3(14b), 3(20), 3(22), 3(23), 3(29) to 3(32), 3(49), 3(55) to 3(57), 3(60), 3(61), 3(63), 3(68); Arts. 4, 5, 6 (incl. 6(3) third subparagraph, profiling), 9, 10, 11, 13, 14, 15, 17, 22 to 27 (incl. 26(11)), 40, 41, 43, 47, 48, 50, 53 (incl. 53(1)(c)), 55, 57, 60, 72, 73, 86; Annexes I, III, IV). Publications Office of the EU (EUR-Lex). 2026-07-27. https://eur-lex.europa.eu/eli/reg/2024/1689/2026-07-27/eng (verified: primary)

Definitions of legal terms paraphrase the cited text, which governs. Dated statements are as of 2026-09-24.

Cite this term

García Aibar, J. (2026). Training, validation and testing data. In AI Governance Engineering: The Thesis & Body of Knowledge (v0.5.0), Glossary. https://doi.org/10.5281/zenodo.22956197. https://aigovernanceengineer.com/glossary/training-validation-and-testing-data. CC BY 4.0

BibTeX

@misc{aige2026trainingvalidationandtestingdata,
  author  = {Jorge García Aibar},
  title   = {{Training, validation and testing data}},
  note    = {Glossary, AI Governance Engineering: The Thesis \& Body of Knowledge, version 0.5.0},
  year    = {2026},
  doi     = {10.5281/zenodo.22956197},
  url     = {https://aigovernanceengineer.com/glossary/training-validation-and-testing-data}
}

Where it is used

Patterns that use this term

Related terms

Sources