With the continuous advancement of data acquisition technology, obtaining large amounts of high-dimensional data containing multiple features has become very easy, such as images and vision data. However, traditional machine learning methods, especially those based on vectors and matrices, face challenges such as the curse of dimensionality, increa...