Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
2
0
5
A global sensitivity analysis of key LLM functions, which bounds the required scale of differentially private noise and derives an upper bound on floating-point error from masking and noise cancellation in the TEE as a function of the privacy parameter epsilon.
Quantum algorithms for differentially private counting queries can actually *amplify* privacy due to inherent randomness, a surprising twist on classical privacy mechanisms.