0bfce127947574733b19da0f30739fcd-AuthorFeedback.pdf
–Neural Information Processing Systems
Our solution exploits the inherent sparseness in the optimization. Each voxel has a very small subset of potential31 fascicles andorientations. Thealgorithm scales wellwithanincrease inthesizeofthetensor. Thegreedy algorithm32 to select orientations does have to iterate over all orientations once, for each voxel, before the optimization is run.33
Neural Information Processing Systems
Feb-11-2026, 10:36:29 GMT