Speaker
Silvia Gazzola
Description
In this poster we present a novel iterative algorithm for low rank tensor completion, tailored to the recovering the missing entries in undersampled X-ray spectromicroscopy data, which are used to study material distributions. Compared to established techniques that rely on data matricizations and low-rank matrix completion, the new method allows the selection of robust sampling patterns, tensor multi-rank and undersampling ratio, while minimising the impact of undersampling on the data analysis. Results obtained on real data will be illustrated.
