Author here - thanks for the feedback everybody! Agree that the weighted version and the skeleton version would be cool to talk about. Perhaps in future :)
yes, this is true. In 2D/3D, sorting into regular space partitions of exponentially decreasing size (e.g. KD-trees) is far more effective - roughly log(N) vs N^rho (rho < 1). In high dimensions, the simple partitioning…
This thread contains many excellent points, and it's true that LSH is no longer SOTA for ANN problems. Generally, LSH indices are much faster to construct but slower to query than other ANN methods (graph-based,…
Author here - thanks for the feedback everybody! Agree that the weighted version and the skeleton version would be cool to talk about. Perhaps in future :)
yes, this is true. In 2D/3D, sorting into regular space partitions of exponentially decreasing size (e.g. KD-trees) is far more effective - roughly log(N) vs N^rho (rho < 1). In high dimensions, the simple partitioning…
This thread contains many excellent points, and it's true that LSH is no longer SOTA for ANN problems. Generally, LSH indices are much faster to construct but slower to query than other ANN methods (graph-based,…