We design differentially private algorithms for the problem of prediction with expert advice under dynamic regret, also known as tracking the best expert. Our work addresses three natural types of adversaries, stochastic with shifting distributions, oblivious, and adaptive, and designs algorithms with sub-linear regret for all three cases. In particular, under a shifting stochastic adversary where the distribution may shift times, we provide an -differentially private algorithm whose expected dynamic regret is at most , where and are the time horizon and number of experts, respectively. For oblivious adversaries, we give a reduction from dynamic regret minimization to static regret minimization, resulting in an upper bound of on the expected dynamic regret, where now denotes the allowable number of switches of the best expert. Finally, similar to static regret, we establish a fundamental separation between oblivious and adaptive adversaries for the dynamic setting: while our algorithms show that sub-linear regret is achievable for oblivious adversaries in the high-privacy regime , we show that any -differentially private algorithm must suffer linear dynamic regret under adaptive adversaries for . Finally, to complement this lower bound, we give an -differentially private algorithm that attains sub-linear dynamic regret under adaptive adversaries whenever .
- ** Work done while at Apple
- † University of Michigan
- ‡ University of Illinois Chicago