BloomFilterPsi
Models & Types · AiDotNet.FederatedLearning.PSI
Implements Bloom filter based probabilistic Private Set Intersection.
Inheritance: PsiBase → BloomFilterPsi
For Beginners
A Bloom filter is like a very compact checklist that can answer “is this item on the list?” with two possible answers:
How It Works
Bloom filter PSI uses a probabilistic data structure to represent one party’s set. The other party queries the Bloom filter for each of its elements. Matches indicate probable intersection membership, with a configurable false-positive rate.
The false-positive rate is configurable. A rate of 0.001 means roughly 1 in 1000 non-matching elements may be incorrectly reported as matching. This is usually acceptable when followed by a verification step.
Complexity: O(n+m) time, O(n) space for the Bloom filter (much smaller than storing the full set). Fastest PSI protocol but probabilistic.
Security note: Standard Bloom filter PSI leaks membership information. For privacy, the Bloom filter should be encrypted or transmitted via secure channel. This implementation simulates the protocol assuming a secure channel.
Reference: Dong et al., “When Private Set Intersection Meets Big Data”, ACM CCS 2013.
Constructors
| Constructor | Summary |
|---|---|
BloomFilterPsi() |
Properties
| Property | Summary |
|---|---|
string ProtocolName { get; } |