Visualizing dissimilarity tensors by polynomials

Multidimensional scaling (MDS) is an approach for presenting data which rather than their actual values, their dissimilarities are accessible. By employing a dissimilarity matrix, it becomes feasible to locate points in $mathbb{R}^k$ in a manner that their pairwise distances closely match the dissimilarities indicated in the matrix. When dealing with data that evolves over discrete time steps, a dissimilarity tensor is used instead of the dissimilarity matrix. In this research paper, we investigate about the problem of MDS for such dissimilarity tensors and aim to identify appropriate polynomials for visualizing the data.