Independent component analysis (ICA) is a statistical method for transforming an observed multidimensional random vector into components that are as independent as possible. In this chapter, we introduce the background information, the theory of ICA, and present several common algorithms such as fast ICA, kernel ICA, and constrained ICA. It is first applied to mineral resources prediction and remote sensing imagery, while traditional methods cannot satisfy the complexity of the spatial data (prospecting geochemistry data, remote sensing data, etc.). In application cases, ICA is applied to analyze the spatial data in some districts of China. The result shows that some independent elements accord with the practical distribution better than conventional methods. Moreover, ICA can get rid of the various kinds of correlations in remote sensing imagery effectively and improve the classification accuracy. However, this method also has some limitations. At last, we list the future research directions of our work.
Deep Learning‐based Classification of Resting‐state fMRI Independent‐component Analysis
