基于神经网络的压缩感知观测序列建模

首先阐述了信号处理的热点———压缩感知技术的基本理论框架,并且分析了压缩感知中观测序列的相关特性,接着将BP神经网络应用于非线性时间序列建模与预测中,相应地提出了加入观测序列建模预测后的CS理论框架,基于此框架,给出了实验仿真的结果,得出结论:利用神经网络对压缩感知观测序列进行建模预测可以进一步减少观测序列的传输量,并且具有很高的预测精度。

Measurements Modeling Based on Neural Network in Compressed Sensing

This paper sets out the hot spot of signal processing.The basic theoretical framework of compressed sensing(CS) is introduced with the analysis of the relevant characteristics of the measurements sequence in CS firstly.Then the BP neural network and its application on the nolinear time sequence modeling are presented with a CS theoretical framework after the modeling and calculating of the measurements sequence.Based on this kind of framework,the experimental results show that it could further reduce the amount of measurements sequence of transport by making use of neural network to model and calculate the measurements sequence,which provides a high precision of prediction.