Classification under local differential privacy
This addresses privacy-preserving classification for data-sensitive applications, but it is incremental as it builds on existing differential privacy frameworks.
The paper tackles binary classification under local differential privacy by proposing a privacy mechanism and constructing a universally consistent classifier in Euclidean spaces, showing that minimax convergence rates of excess risk are slower than with non-private data.
We consider the binary classification problem in a setup that preserves the privacy of the original sample. We provide a privacy mechanism that is locally differentially private and then construct a classifier based on the private sample that is universally consistent in Euclidean spaces. Under stronger assumptions, we establish the minimax rates of convergence of the excess risk and see that they are slower than in the case when the original sample is available.