The topological-ordering property of the SOM algorithm may be used to form an abstract two-dimensional representation of a high-dimensional input space to investigate this form of a representation, consider a two-dimensional lattice consisting of a 10 X 10 network of neurons that is trained with an input consisting of four Gaussian clouds in an eight-dimensional input space All the clouds
have unit variance, but different centers. The centers are located at the points (0, 0, 0, ...,0), (4, 0, 0, ..., 0), (4, 4, 0, ..., 0), and (0, 4, 0, ..., 0). Compute the map produced by the SOM algorithm, with each neuron in the map being labeled with the particular class most frequently represented by the input points around it.