Nearest neighbor search has a long history in theoretical computer science, and in the past decade has seen an explosion of usage. This has been primarily driven by embedding models that represent complex data—text, images, and video—as high-dimensional vectors, shining a new spotlight on vector retrieval. I’ll discuss ongoing work with Ashwin Padaki and Sanjeev Khanna exploring new takes on this problem. My goal is to show you some new directions from the theoretical computer science perspective, and a few plots (!) comparing our algorithmic ideas with modern benchmarks.

Zoom: https://upenn.zoom.us/j/92067813194