English

Evaluating Multiple Guesses by an Adversary via a Tunable Loss Function

Information Theory 2021-08-20 v1 math.IT

Abstract

We consider a problem of guessing, wherein an adversary is interested in knowing the value of the realization of a discrete random variable XX on observing another correlated random variable YY. The adversary can make multiple (say, kk) guesses. The adversary's guessing strategy is assumed to minimize α\alpha-loss, a class of tunable loss functions parameterized by α\alpha. It has been shown before that this loss function captures well known loss functions including the exponential loss (α=1/2\alpha=1/2), the log-loss (α=1\alpha=1) and the 00-11 loss (α=\alpha=\infty). We completely characterize the optimal adversarial strategy and the resulting expected α\alpha-loss, thereby recovering known results for α=\alpha=\infty. We define an information leakage measure from the kk-guesses setup and derive a condition under which the leakage is unchanged from a single guess.

Keywords

Cite

@article{arxiv.2108.08774,
  title  = {Evaluating Multiple Guesses by an Adversary via a Tunable Loss Function},
  author = {Gowtham R. Kurri and Oliver Kosut and Lalitha Sankar},
  journal= {arXiv preprint arXiv:2108.08774},
  year   = {2021}
}

Comments

6 pages

R2 v1 2026-06-24T05:15:32.099Z