Mean Shift Clustering allows for a wide range of cluster sizes and forms by searching for the modes (peaks) of data density without requiring a set number of clusters.

In order to divide data into clusters, spectral clustering uses the eigenvalues and eigenvectors of a similarity matrix in a graph representation of the data. This allows it to handle non-convex clusters for applications like community discovery and picture segmentation.

Since fuzzy clustering, and more especially fuzzy C-Means, allows data points to belong to several clusters to differing degrees, it is a suitable option when it is difficult to identify definite cluster borders.