Hey everyone! I'm super excited to present a paper summary of HNSW-FINGER! HNSW-FINGER presents a clever technique to approximate distances by projecting the query and candidate nodes onto the center node being explored! The video covers all sorts of details around how this is computed and the memory cost of doing so, as well as the experimental results! More than happy to answer any questions or discuss any ideas you have about this paper! Thank you so much for watching!
Paper Link: assets.amazon....
Chapters
0:00 Introduction
1:20 2 Minute Overview
4:45 Presentation Topics
6:48 HNSW Search
10:15 Approximating L2 Distance
20:25 Memory Cost
23:56 Distribution Matching
24:55 Results
27:18 My Takeaways