A question about the ONNG paper

yahoojapan / NGT

Nearest Neighbor Search with Neighborhood Graph and Tree for High-dimensional Data

Apache License 2.0

1.26k stars 115 forks source link

A question about the ONNG paper #101

Closed coraler closed 3 years ago

coraler commented 3 years ago

Your work was great and I learned a lot！ When reading ONNG paper, the value of e_o used by algorithm 3 when calling algorithm 2 is 0, can you explain the reason? Wouldn't algorithm 2 never enter the loop, or should it be e_o? It would be a great help if you could answer this question,thanks!

masajiro commented 3 years ago

Thank you for your interest in NGT. 0 of ConstructAdjustedGraph(G, 0, e) In Algorithm 3 is to build TKNNG that is a transposed graph of KNNG, and there is a mistake In Algorithm 2.

5: while S ≠ ∅ and p ≤ e_o and p ≤ e_i → 5: while S ≠ ∅ and (p ≤ e_o or p ≤ e_i)