Deep Deterministic Independent Component Analysis for Hyperspectral Unmixing
This work addresses hyperspectral unmixing for remote sensing applications, presenting a novel method that challenges existing assumptions, though it appears incremental in advancing ICA techniques.
The authors tackled the problem of hyperspectral unmixing by developing a new neural network-based independent component analysis method that minimizes dependence among extracted components, achieving results that refute prior claims about ICA's ineffectiveness in this domain.
We develop a new neural network based independent component analysis (ICA) method by directly minimizing the dependence amongst all extracted components. Using the matrix-based R{é}nyi's $α$-order entropy functional, our network can be directly optimized by stochastic gradient descent (SGD), without any variational approximation or adversarial training. As a solid application, we evaluate our ICA in the problem of hyperspectral unmixing (HU) and refute a statement that "\emph{ICA does not play a role in unmixing hyperspectral data}", which was initially suggested by \cite{nascimento2005does}. Code and additional remarks of our DDICA is available at https://github.com/hongmingli1995/DDICA.