Local Differential Privacy
A variant of differential privacy that requires each user to add noise to data locally before uploading it, eliminating the need for a trusted central server.
Timeline (last 90 days)
Future directions include relaxing the oblivious assumption, studying more general privacy models (such as local differential privacy), and extending the technique to other online learning problems (such as bandit feedback).
Randomized response is a classical local differential privacy mechanism whose basic idea is to answer truthfully with a certain probability and give a random answer with another probability, thereby concealing true information at the individual level.
The study is still based on the BTL model and the linear reward assumption, and future work needs to extend to more complex preference models (such as deep neural network reward models) and broader privacy definitions (such as combinations of local and centralized differential privacy).
All Facts (3)
Randomized response is a classical local differential privacy mechanism whose basic idea is to answer truthfully with a certain probability and give a random answer with another probability, thereby concealing true information at the individual level.
50%UnverifiedThe study is still based on the BTL model and the linear reward assumption, and future work needs to extend to more complex preference models (such as deep neural network reward models) and broader privacy definitions (such as combinations of local and centralized differential privacy).
50%UnverifiedFuture directions include relaxing the oblivious assumption, studying more general privacy models (such as local differential privacy), and extending the technique to other online learning problems (such as bandit feedback).
50%