基于局部傅里叶变换数据的重构算法的研究,是一种比有效快速的重构算法-New features: Switched to complex computation, fully compatible with complex data Replaced penalty parameter continuation by R. Glowinski and P.L. Tallec s alternating direction method (ADM or ADMM) New

训练集/测试集产生 load spectra_data.mat 随机产生训练集和测试集 temp = randperm(size(NIR,1)) 训练集——50个样本 P_train = NIR(temp(1:50),:) T_train = octane(temp(1:50),:) 测试集——10个样本 P_test = NIR(temp(51:end),:) T_test = octane(temp(51:end),:) N = si