Differential privacy and the protected unit
Research context
postgraduate concept · established · As of 2026-10-10
A randomized mechanism is (epsilon,delta)-differentially private if every neighboring dataset pair and every output event obey Pr[M(D) in S]≤exp(epsilon)Pr[M(D-prime) in S]+delta.
Key knowledge
A randomized mechanism is (epsilon,delta)-differentially private if every neighboring dataset pair and every output event obey Pr[M(D) in S]≤exp(epsilon)Pr[M(D-prime) in S]+delta. The adjacency relation defines whether one record, person or other unit is protected; add/remove and replacement conventions change sensitivity. DP limits output-distribution change rather than guaranteeing anonymity, truthful data or absence of any inference about a population.
Values and equations
Pr[M(D)∈S]≤e^epsilon Pr[M(D')∈S]+delta
- D,D'
- Datasets adjacent under the declared protected-unit relation
- M
- Randomized release mechanism
- epsilon,delta
- Privacy parameters
- S
- Measurable output event
Conditions and scope
- Adjacency is symmetric and explicitly declared; the guarantee quantifies over all adjacent inputs and measurable output events.
- epsilon≥0 and 0≤delta<1; data-dependent releases and preprocessing belong to the protected mechanism.
Sources · 1
References
Sources retain their recorded scope. A source link alone does not establish verification of every statement.
Evidence and review
Author checked this English statement, displayed equation where present and stated hypotheses against these source locations: Definition 2.4, postprocessing, composition, Laplace mechanism and advanced composition Theorem 3.20. The check is bounded to the recorded claims; no complete proof reconstruction, software proof replay or universal literature completeness is claimed.
2026-10-10 · Author source-body/statement/equation/domain check, with graph associations scoped individually. Separate independent content/source/inventory review is required before branch closure.
Entry details
Permanent link to this entry ↗- Record identity
- research-info-differential-privacy-adjacency
- Revision
- research.1
- Structured reference record
- Postgraduate & research reference