Canonical Polyadic

Canonical Polyadic (CP) decomposition is a tensor factorization technique aiming to represent a multi-dimensional array as a sum of rank-one tensors, facilitating data analysis and dimensionality reduction. Current research focuses on improving CP's robustness and efficiency, particularly through extensions like Block-Term Decomposition (BTD) which offers a balance between CP and Tucker decompositions, and probabilistic approaches that incorporate uncertainty. These advancements are impacting diverse fields, including signal processing (e.g., blind source separation), machine learning (e.g., neural network compression), and data analysis of large-scale dynamic networks, by enabling more efficient and accurate modeling of complex datasets.