arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1905.05313cond-mat.dis-nncond-mat.stat-mechcs.ITcs.LGmath.IT
收藏

高维估计的广义近似调查传播

Generalized Approximate Survey Propagation for High-Dimensional Estimation

Italian Institute for Genomic Medicine(意大利基因组医学研究所);Microsoft Research New England(微软新英格兰研究院);John A. Paulson School of Engineering and Applied Sciences, Harvard University(哈佛大学约翰·A·保尔森工程与应用科学学院);Bocconi Institute for Data Science and Analytics, Bocconi University(博科尼大学博科尼数据科学与分析研究所)

Luca Saglietti, Yue M. Lu, Carlo Lucibello

摘要:

在广义线性估计(GLE)问题中,我们试图估计通过线性变换观察到的信号,然后是一个分量,可能是非线性和噪声信道。在贝叶斯最优设置,广义近似消息传递(GAMP)是已知的GLE实现最佳性能。然而,它的性能可以显着降低,只要有一个假设和真实的生成模型之间的不匹配,在实践中经常遇到的情况。在本文中,我们提出了一种新的算法,命名为广义近似调查传播(GASP),解决GLE在存在的先验或模型错误的规格。作为一个典型的例子,我们考虑的相位恢复问题,在那里我们表明,GASP优于相应的GAMP,降低重建阈值,并为某些参数的选择,接近贝叶斯最优性能。此外,我们提出了一组状态演化方程,准确地描述了GASP的动力学在高维极限。

英文摘要:

In Generalized Linear Estimation (GLE) problems, we seek to estimate a signal that is observed through a linear transform followed by a component-wise, possibly nonlinear and noisy, channel. In the Bayesian optimal setting, Generalized Approximate Message Passing (GAMP) is known to achieve optimal performance for GLE. However, its performance can significantly degrade whenever there is a mismatch between the assumed and the true generative model, a situation frequently encountered in practice. In this paper, we propose a new algorithm, named Generalized Approximate Survey Propagation (GASP), for solving GLE in the presence of prior or model mis-specifications. As a prototypical example, we consider the phase retrieval problem, where we show that GASP outperforms the corresponding GAMP, reducing the reconstruction threshold and, for certain choices of its parameters, approaching Bayesian optimal performance. Furthermore, we present a set of State Evolution equations that exactly characterize the dynamics of GASP in the high-dimensional limit.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑