English

Parametric Bilinear Generalized Approximate Message Passing

Information Theory 2016-05-25 v2 math.IT

Abstract

We propose a scheme to estimate the parameters bib_i and cjc_j of the bilinear form zm=i,jbizm(i,j)cjz_m=\sum_{i,j} b_i z_m^{(i,j)} c_j from noisy measurements {ym}m=1M\{y_m\}_{m=1}^M, where ymy_m and zmz_m are related through an arbitrary likelihood function and zm(i,j)z_m^{(i,j)} are known. Our scheme is based on generalized approximate message passing (G-AMP): it treats bib_i and cjc_j as random variables and zm(i,j)z_m^{(i,j)} as an i.i.d.\ Gaussian 3-way tensor in order to derive a tractable simplification of the sum-product algorithm in the large-system limit. It generalizes previous instances of bilinear G-AMP, such as those that estimate matrices B\boldsymbol{B} and C\boldsymbol{C} from a noisy measurement of Z=BC\boldsymbol{Z}=\boldsymbol{BC}, allowing the application of AMP methods to problems such as self-calibration, blind deconvolution, and matrix compressive sensing. Numerical experiments confirm the accuracy and computational efficiency of the proposed approach.

Keywords

Cite

@article{arxiv.1508.07575,
  title  = {Parametric Bilinear Generalized Approximate Message Passing},
  author = {Jason T. Parker and Philip Schniter},
  journal= {arXiv preprint arXiv:1508.07575},
  year   = {2016}
}
R2 v1 2026-06-22T10:44:37.080Z