Using LMS trees for imaging processing
Terence D. Sanger · 1991
LMS (least mean squares) trees are applied to image coding and image filtering tasks. The ability of these trees to handle high-dimensional input makes them appropriate for image processing, in which large numbers of inputs may be required even for simple tasks. A tree pruning method for increasing the forward propagation speed of the networks is proposed based on recursively eliminating subtrees with small weights.>