针对现有的显著性检测算法检测目标类型单一、通用性差的问题,提出一种基于无监督栈式降噪自编码网络的显著性检测算法.该算法利用无监督栈式降噪自编码网络(Stacked Denoising Auto Encoder,SDAE)在多个尺度对原始图像进行稀疏重构,将...针对现有的显著性检测算法检测目标类型单一、通用性差的问题,提出一种基于无监督栈式降噪自编码网络的显著性检测算法.该算法利用无监督栈式降噪自编码网络(Stacked Denoising Auto Encoder,SDAE)在多个尺度对原始图像进行稀疏重构,将原始图像与SDAE网络重构图像之间的差作为显著图,二值化后的显著图作为显著性目标检测结果.在SDAE网络训练过程中,将原始图像作为原始数据,网络重构的图像作为观察数据.为了提升网络训练效率,首先利用无监督逐层贪婪方法训练同结构的深度信念网络(Deep Belief Network,DBN),将训练得到的DBN网络参数设为SDAE网络的初始参数,再计算原始数据与观察数据之间的互信息作为网络收敛代价,利用反向传播进行网络参数微调.实验表明,该网络模型可以完成多类型目标的显著性检测,具有通用性好,准确度高等优点.展开更多
Large low-field magnetoresistance(LFMR)of∆p/pH=40%is obtained at 163 K within±3000 Oe in a polycrystalline La_(0.7)Ca_(0.3)MnO_(3)(LCMO)films prepared by pulsed laser deposition.The ratio∆p/pH increases almost li...Large low-field magnetoresistance(LFMR)of∆p/pH=40%is obtained at 163 K within±3000 Oe in a polycrystalline La_(0.7)Ca_(0.3)MnO_(3)(LCMO)films prepared by pulsed laser deposition.The ratio∆p/pH increases almost linearly with H in the measuring magnetic field range and the peak of LFMR ratios is observed near the resistivity peak temperature,which is different from the observation of spin-polarized intergrain tunnels.This experiment suggests that grain boundaries have large influences on the LFMR.The results indicate that a suitable film process can be used to increase the LFMR of LCMO films.展开更多
基金Supported by the National Natural Science Foundation of China under Grant No.19674001.
文摘Large low-field magnetoresistance(LFMR)of∆p/pH=40%is obtained at 163 K within±3000 Oe in a polycrystalline La_(0.7)Ca_(0.3)MnO_(3)(LCMO)films prepared by pulsed laser deposition.The ratio∆p/pH increases almost linearly with H in the measuring magnetic field range and the peak of LFMR ratios is observed near the resistivity peak temperature,which is different from the observation of spin-polarized intergrain tunnels.This experiment suggests that grain boundaries have large influences on the LFMR.The results indicate that a suitable film process can be used to increase the LFMR of LCMO films.
