Glossary

Membership inference attack (MIA)

An attack that builds shadow models and aggregates prediction confidence differences between member and non-member samples to estimate membership likelihood.

Source: Collaborative Application Technology for Financial Risk Control Models under Privacy Preservation Constraints (Academic research) · p.6

Have a question about Membership inference attack (MIA)? Ask the AI, answers cite the source.