Terms related to neural networks · 3.4.4
deep learning
an approach that builds rich hierarchical representations by training neural networks that have many hidden layers
What this term settles
The definition is deliberately narrow. It tells the reader what counts as deep learning and leaves the engineering choices outside the word.
Who it binds
Nothing here creates an obligation by itself, and the obligation for deep learning arrives when the firm writes the term into a control, a specification or a purchase order.
What to do in your practice
Take the firm's own definition of deep learning, put it beside this one and mark the differences. Each difference is either a deliberate local choice or an inconsistency to fix.
Evidence an auditor or regulator asks for
- A controlled vocabulary entry for deep learning with the citation 3.4.4, an owner and a review date.An auditor can check that deep learning is controlled rather than assumed.
- A model architecture note that names the network form and the layers used. The row names deep learning.Shows deep learning in use in the artefact the firm relies on, not only in a glossary.
- A test record that shows the behaviour the architecture was chosen for. The record carries the deep learning entry and its date.The architecture note uses the network name as a marketing label with no layer or training detail behind it. The gap is tested against deep learning.
Common gaps this page catches
- deep learning is used in the network architecture note without a written definition, so two teams can mean different things.
- The entry for deep learning is reviewed when someone remembers, not on a cadence, so it drifts from the artefact it describes.
What to prepare before the amendment lands
Put deep learning on the agenda of the next AI governance meeting with two questions: where is it used, and what would show it is used correctly? Record the answers as actions.
How to test it here
Take one live artefact rather than a template, find the sentence or field where deep learning is used, and put this page beside it. Record the answer, the artefact and the date, because that record is the evidence and a glossary entry on its own is only an intention.
What this page does not claim
The page does not reproduce the standard, certify the firm or predict the wording the amendment will use; it states deep learning at the claim level "definition" and stops where the public record stops.
Common questions
- What does deep learning mean in ISO 22989?
- an approach that builds rich hierarchical representations by training neural networks that have many hidden layers
- What evidence should we keep for 3.4.4?
- A controlled vocabulary entry for deep learning with the citation 3.4.4, an owner and a review date. A model architecture note that names the network form and the layers used. The row names deep learning.
- Is deep learning a requirement?
- No: deep learning is a definition or clause of a terminology standard, and it becomes binding only when a firm writes it into a policy, a contract or a procedure, or when another standard incorporates it.
Related units
Source. ISO/IEC 22989:2022, Clause 3.4.4, Terms and definitions. Open the ISO public entry. Retrieved 21 October 2026.
https://www.iso.org/obp/ui/en/#iso:std:iso-iec:22989:ed-1:v1:en
ISO catalogue record 88145, the amendment status, stage and dates. Catalogue record 74296, the base standard.
Claim level: definition. The definition is in the public preview. This page states it in our own words and does not reproduce the text.
Last updated: 2026-10-21 · Markdown version · Check your vocabulary