Poisson Sampling
A subsampling method commonly used in differential privacy in which each data point is independently selected into a batch with a fixed probability, thereby providing privacy amplification.
Timeline (last 90 days)
This privacy analysis framework uniformly covers both Poisson sampling and random allocation sampling schemes, and supports efficient and relatively tight privacy accounting.
The paper empirically evaluates Poisson sampling with truncated batches, comparing against Ganesh (2025)'s direct accounting method for truncation, and the results show that this mechanism performs better in the privacy-efficiency trade-off.
The paper's analysis targets Poisson sampling and random allocation, and the applicability to other sampling schemes is outside the scope of discussion.
All Facts (3)
This privacy analysis framework uniformly covers both Poisson sampling and random allocation sampling schemes, and supports efficient and relatively tight privacy accounting.
50%UnverifiedThe paper empirically evaluates Poisson sampling with truncated batches, comparing against Ganesh (2025)'s direct accounting method for truncation, and the results show that this mechanism performs better in the privacy-efficiency trade-off.
50%UnverifiedThe paper's analysis targets Poisson sampling and random allocation, and the applicability to other sampling schemes is outside the scope of discussion.
50%