Latest revision as of 00:24, 20 January 2026

Backdoor pattern

A transformation or insertion applied to a data sample that triggers an adversary-specified behaviour in a model that has been subject to a backdoor poisoning attack. For example, in computer vision, an adversary could poison a model such that the insertion of a square of white pixels induces a desired target label.

Source: NIST AI 100-2e2025 | Category:

Revision as of 00:12, 15 January 2026 view source imported>Unknown user No edit summary	Latest revision as of 00:24, 20 January 2026 view source imported>Unknown user No edit summary
(2 intermediate revisions by the same user not shown)
(No difference)