Dataset Auditing
A technique in model governance for determining whether a model was trained or fine-tuned on a specific dataset; it is divided into passive auditing and active auditing.
Timeline (last 90 days)
SpectralCanary is a proactive dataset auditing method that blends a fixed canary image into every dataset image using a smooth low-pass mask, requiring no additional generation or optimization process.
In dataset auditing scenarios, a low false positive rate (FPR) is crucial, because falsely accusing a model of using a certain dataset may bring legal or reputational risks.
All Facts (2)
SpectralCanary is a proactive dataset auditing method that blends a fixed canary image into every dataset image using a smooth low-pass mask, requiring no additional generation or optimization process.
50%UnverifiedIn dataset auditing scenarios, a low false positive rate (FPR) is crucial, because falsely accusing a model of using a certain dataset may bring legal or reputational risks.
50%