Revision as of 02:53, 15 January 2026

Shadow model

A model that imitates the behavior of the target model. The training datasets and the truth about membership in these datasets are known for these models. Typically, the attack model is trained on the labeled inputs and outputs of the shadow model.

Source: NIST AI 100-2e2025 | Category:

Revision as of 01:42, 15 January 2026 view source imported>Unknown user No edit summary ← Older edit	Revision as of 02:53, 15 January 2026 view source imported>Unknown user No edit summary Newer edit →
(No difference)