FAQs
Differential Privacy Basics
What is the Formal Definition of Differential Privacy?
FAQs
Differential Privacy Basics
What is the Formal Definition of Differential Privacy?

Differential privacy is defined as a property of algorithms used for data analysis. An algorithm provides ε-differential privacy if, for any two datasets differing in only one individual's data, and for every possible outcome, the probability of the outcome occurring with one dataset is at most e^ε times the probability with the other dataset. Here, ε is a small positive number that quantifies privacy loss, with smaller values indicating stronger privacy. When ε is zero the querier learns nothing from the dataset, when it is infinite they would learn every record in the dataset with absolute assurance. Privacy protection is a sliding scale between these two extremes. This definition ensures that the algorithm's output is not significantly influenced by the presence or absence of any single individual's data, providing a strong guarantee of privacy.

Curious about implementing DP into your workflow?

Join Antigranular

Got questions about differential privacy?

Ask us on Discord

Want to check out articles on similar topics?

Read the blog

2024 Oblivious Software Ltd. All rights reserved.