We study the median slope selection problem in the oblivious RAM model. In this model memory accesses have to be independent of the data processed, i.e., an adversary cannot use observed access patterns to derive additional information about the input. We show how to modify the randomized algorithm of Matou\v{s}ek (1991) to obtain an oblivious version with $\mathcal{O}(n \log^2 n)$ expected time for $n$ points in $\mathbb{R}^2$. This complexity matches a theoretical upper bound that can be obtained through general oblivious transformation. In addition, results from a proof-of-concept implementation show that our algorithm is also practically efficient.
翻译:我们研究了隐蔽的内存模型的中位坡度选择问题。 在这个模型中, 内存访问必须独立于所处理的数据, 也就是说, 对手不能使用观察到的存取模式来获取关于输入的补充信息。 我们展示了如何修改 Matu\ v{s}ek 的随机算法( 1991), 以获得一个以$\ mathcal{O}( n\log2n) 计算的未识别版本, 以$\ mathbb{R}2$ 。 这种复杂性与通过一般的模糊变换可以获得的理论上限相匹配。 此外, 验证概念执行的结果显示, 我们的算法也非常有效 。</s>