基于生成模型的块稀疏偏差建模  

Block Sparse Deviation Modeling Through Generative Models

作　　者：斯那雨追 余鹏彬 王建军[1] SINA Yuzhui;YU Pengbin;WANG Jianjun(School of Mathematic and Statistics,Southwest University,Chongqing 400715,China)

机构地区：[1]西南大学数学与统计学院,重庆400715

出　　处：《西南师范大学学报（自然科学版）》2023年第1期81-89,共9页Journal of Southwest China Normal University(Natural Science Edition)

基　　金：国家自然科学基金项目(12071380)。

摘　　要：研究了基于生成模型的块稀疏偏差模型(Block Sparse-Gen).依靠块RIP条件和块REC条件,从理论上给出了最优解码的重构误差以及高概率恢复生成函数的测量次数.实验数值也验证了Block Sparse-Gen的有效性.In this paper, we concentrate on studying the compressed sensing block sparse deviation model based on the generative model(Block Sparse-Gen). Relying on the Block RIP condition and the Block REC condition, we theoretically gave the reconstruction error for the optimal decoding and the number of measurements for the high probability recovery generating function. Moreover, the experimental values verifies the effectiveness of Block Sparse-Gen.

关 键 词：生成模型 块稀疏 压缩感知 块稀疏偏差模型 

分 类 号：TN911.73[电子电信—通信与信息系统]