Papers · Proving properties · FairZK
FairZK
FairZK: A Scalable System to Prove Machine Learning Fairness in Zero-Knowledge
Reported benchmarks
| model | params | proving time s | speedup vs prior | note |
|---|---|---|---|---|
| DNN | 47M | 343 | 3.1–1789 | First system to prove fairness at 47M parameters. |
Notes
The scalability unlock: derives fairness bounds from MODEL PARAMETERS plus aggregated input statistics, rather than by proving inference over a specific dataset. That is why it reaches 47M params where inference-based fairness proofs cannot. Yupeng Zhang also co-authored zkCNN and DeepProve.
Our reading
Citation neighbourhood
Builds on
- none recorded
Cited by, in this corpus
- none recorded
Edges are a proxy: paper A's text mentions B anywhere (body or bibliography). See the full graph.
Discussed in
A claim about a model, not about a computationThe prior surveys, and where we disagree with themEvery relaxation, and what an adversary can still do
Other recorded fields
property_proven: Group fairness of logistic regression and DNNs