Kernelize

Kernelization transforms bilinear data—typically arising from exponential decays—into a richer, trilinear representation by convolving signals with a set of normalized kernels. These kernels capture essential temporal features, such as impulsive events and gradual decays, expanding the data into an additional dimension. The resulting trilinear structure is especially useful in applications like multiexponential signal analysis, where decomposition of complex, overlapping signals is critical.