brainspace.gradient.utils.dominant_set

brainspace.gradient.utils.dominant_set(s, k, is_thresh=False, norm=False, copy=True, as_sparse=True)[source]

Keep the largest elements for each row. Zero-out the rest.

Parameters:
  • s (2D ndarray) – Similarity/affinity matrix.
  • k (int or float) – If int, keep top k elements for each row. If float, keep top 100*k percent of elements. When float, must be in range (0, 1).
  • is_thresh (bool, optional) – If True, k is used as threshold. Keep elements greater than k. Default is False.
  • norm (bool, optional) – If True, normalize rows. Default is False.
  • copy (bool, optional) – If True, make a copy of the input array. Otherwise, work on original array. Default is True.
  • as_sparse (bool, optional) – If True, return a sparse matrix. Otherwise, return the same type of the input array. Default is True.
Returns:

output (2D ndarray or sparse matrix) – Dominant set.