Bloom filter: space-efficient probabilistic set membership. k hash functions → k bit positions. False positive rate: (1-e^(-kn/m))^k. No false negatives. Delete not supported. Counting Bloom filter: allows deletion with counter array. Cuckoo filter: supports deletion, better cache performance. HyperLogLog: cardinality estimation with O(log log n) space. Count-Min Sketch: frequency estimation. MinHash: Jaccard similarity estimation. Applications: CDN cache routing, database query...
- data-structures
- probabilistic