Spiegelberg Jakob, Rusz Ján, Pelckmans Kristiaan
Department of Physics and Astronomy, Uppsala University, Box 516, S-751 20 Uppsala, Sweden.
Department of Information Technology, Uppsala University, Box 337, S-751 05 Uppsala, Sweden.
Ultramicroscopy. 2017 Apr;175:36-45. doi: 10.1016/j.ultramic.2016.12.025. Epub 2017 Jan 11.
A selection of tensor decomposition techniques is presented for the detection of weak signals in electron energy loss spectroscopy (EELS) data. The focus of the analysis lies on the correct representation of the simulated spatial structure. An analysis scheme for EEL spectra combining two-dimensional and n-way decomposition methods is proposed. In particular, the performance of robust principal component analysis (ROBPCA), Tucker Decompositions using orthogonality constraints (Multilinear Singular Value Decomposition (MLSVD)) and Tucker decomposition without imposed constraints, canonical polyadic decomposition (CPD) and block term decompositions (BTD) on synthetic as well as experimental data is examined.
本文介绍了一系列张量分解技术，用于检测电子能量损失谱（EELS）数据中的微弱信号。分析的重点在于模拟空间结构的正确表示。提出了一种结合二维和n路分解方法的EEL光谱分析方案。具体而言，研究了稳健主成分分析（ROBPCA）、使用正交性约束的塔克分解（多线性奇异值分解（MLSVD））和无约束的塔克分解、典范多向分解（CPD）以及合成数据和实验数据上的块项分解（BTD）的性能。
