当前位置: X-MOL 学术Theor. Comput. Sci. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
An Improved Algorithm for Learning Sparse Parities in the Presence of Noise
Theoretical Computer Science ( IF 1.1 ) Pub Date : 2021-04-30 , DOI: 10.1016/j.tcs.2021.04.026
Di Yan , Yu Yu , Hanlin Liu , Shuoyao Zhao , Jiang Zhang

We revisit the Learning Sparse Parities with Noise (LSPN) problem on k out of n variables for kn, and present the following findings.

1.

For true parity size k=nu for any 0<u<1, and noise rate η<1/2, the first algorithm solves the (n,k,η)-LSPN problem with constant probability and time/sample complexity n(1u+o(1))k(1/2η)2.

2.

For any 1/2<c1<1, k=o(ηn/logn), and ηnc1/4, our second algorithm solves the (n,k,η)-LSPN problem with constant probability and time/sample complexity n2(1c1+o(1))k.

3.

We show a “win-win” result about reducing the number of samples. If there is an algorithm that solves (n,k,η)-LSPN problem with probability Ω(1), time/sample complexity nO(k) for k=o(n1c), any noise rate η=n12c/3 and 1/2c<1. Then, either there exists an algorithm that solves the (n,k,μ)-LSPN problem under lower noise rate μ=nc/3 using only 2n samples, or there exists an algorithm that solves the (n,k,μ)-LSPN problem for a much larger k=n1c with probability nO(k)/poly(n), and time complexity poly(n)nO(k), using only n samples.

Our algorithms are simple in concept by combining a few basic techniques such as majority voting, reduction from the LSPN problem to its decisional variant, Goldreich-Levin list decoding, and computational sample amplification.



中文翻译:

噪声存在下稀疏奇偶校验的一种改进算法

我们重温学习稀疏平价与噪声(LSPN)问题ķñ变量ķñ,并提出以下发现。

1。

真正的奇偶校验大小 ķ=ñü 对于任何 0<ü<1个和噪音率 η<1个/2个,第一种算法以恒定的概率和时间/样本复杂度解决了(nkη)-LSPN问题ñ1个-ü+Ø1个ķ1个/2个-η2个

2。

对于任何 1个/2个<C1个<1个ķ=Øηñ/日志ñ, 和 ηñ-C1个/4,我们的第二种算法以恒定的概率和时间/样本复杂度解决了(nkη)-LSPN问题ñ2个1个-C1个+Ø1个ķ

3。

我们展示了减少样本数量的“双赢”结果。如果有一种算法可以解决ñķη-LSPN问题的概率 Ω1个,时间/样本复杂度 ñØķ 为了 ķ=Øñ1个-C,任何噪音率 η=ñ1个-2个C/31个/2个C<1个。然后,要么存在一种可以解决以下问题的算法ñķμ-低噪声率下的LSPN问题 μ=ñ-C/3仅使用2 n个样本,或者存在一种算法可以解决ñķμ-LSPN问题更大 ķ=ñ1个-C 很有可能 ñ-Øķ/ñ和时间复杂度 ññØķ,仅使用n个样本。

我们的算法在概念上很简单,它结合了一些基本技术,例如多数表决,从LSPN问题减少到其决策变型,Goldreich-Levin列表解码和计算样本放大。

更新日期:2021-04-30
down
wechat
bug