The core concepts in differential privacy revolve around the idea of providing privacy guarantees through controlled random noise. Key definitions include:

- Neighboring Datasets: Two datasets are considered neighbouring if they differ by only one individual's data. This concept is central to defining how much an individual's data can influence the overall analysis.

- Randomised Mechanism: A fundamental tool in differential privacy, a randomised mechanism applies a random process (often involving noise addition) to the data or query responses, ensuring that the output does not depend significantly on any single individual's data.

- Privacy Loss Random Variable: This concept measures the change in the probability distribution of the mechanism's output when an individual's data is modified, providing a quantifiable measure of privacy risk.