← Back
Home › Systems › Security and identity

differential-privacy

differentially private analytics

RAPPOR, Apple DP

SystemTechniques · 4Primitives · 16 differential-privacy Dp sgdml-training · medium: Differentially private SGD: per-example gradient clipping plus Gaussian noise, with privacy accountingDifferential privacysecurity · medium: Calibrated noise (Laplace/Gaussian) and privacy budgets for aggregate releasesPrivacy budget accountingsecurity · medium: Tracking cumulative epsilon/delta under composition (basic, advanced, Renyi/moments accountant)Secure aggregationsecurity · medium: Summing client vectors so the server sees only the total: pairwise masks, secret-shared seeds, dropout recoveryCsprngsecurity · low: OS-backed cryptographically secure random numbers (getrandom), and when a fast PRNG is not enoughFloating-point errornumerics · low: How IEEE 754 rounding error accumulates and how to contain it: compensated (Kahan/Neumaier) and pairwise summation, error-free transformations with FMA, ULP comparisons.›Bloom filterdata-structures · low: Bit array answering "definitely not present" or "probably present" in about 10 bits per key for 1 % false positives; inserts are cheap, deletes impossible.›Count-min sketchdata-structures · low: d rows of w counters estimate any item's frequency in a stream, overestimating by at most ε·N with probability 1-δ, in fixed tiny memory.›HyperLogLogdata-structures · low: Counts distinct elements in kilobytes with about 1 % error by tracking the longest run of leading zero bits in hashed values per bucket.›Fast walsh hadamard transformnumerics · low: O(n log n) Walsh-Hadamard transform used in sketches, randomized projections and local DPRandomized responsesecurity · low: Local DP by answering truthfully with probability p, else randomly; unbiased frequency estimationGradient clippingml-training · low: Clipping gradients by global norm (or value) to prevent loss spikes; interaction with mixed precisionSGD with momentumml-training · low: Gradient descent plus a decaying velocity buffer that averages minibatch noise and speeds up progress along shallow, consistent directions.›Conditional writesdistributed · low: Put-if-absent / if-match compare-and-swap on objects or keys as a commit primitiveFinite field arithmeticnumerics · low: Prime-field and extension-field arithmetic (Montgomery form, Goldilocks, BabyBear), vectorizedPRNGs (PCG, xoshiro)numerics · low: Small, fast, statistically strong non-cryptographic generators: PCG32 and xoshiro256**, with seeding, parallel streams and unbiased ranges.›ECDH key exchange (X25519, P-256)security · low: Each side multiplies the other's public curve point by its own scalar and both get the same point; always validate the peer key and run the result through a KDF.›Shamir secret sharing (k-of-n)security · low: Split a secret into n shares so any k reconstruct it and any k-1 reveal nothing at all, using a random polynomial over a finite field.›d-ary heapdata-structures · low: Array heap with d children per node (d = 4 typical): half the binary heap's depth, siblings in one cache line, faster priority queues.›Fast hash functionsdata-structures · low: Non-cryptographic hashes built from 64×64→128-bit multiplies and xor-shifts; pick by key size, trust level and whether the hash is persisted.›

Similar systems

Share the most building blocks with this one.

federated-learning 20 %