Learning nonlinearities in the decoupling problem with structured CPD

Decoupling of multivariate functions is an important problem in block-structured system identification. In the literature, different tensor-based solutions have been proposed to solve this using a canonical polyadic decomposition (CPD) Jacobian tensor. recent work, it has add polynomial constraint on model, which leads constrained CPD model with two factors depending nonlinearly each other. we are interested estimation static nonlinearities constraints. Using reformulation as structured CPD, propose alternating least squares algorithm rank-one terms. Numerical simulations show performance and usefulness method compared competing approach.