Advancements in Approximate Nearest Neighbors Search: Algorithmic Approaches for Efficient Indexing and Retrieval

Speaker

Silvio Martinico (University of Pisa)

– Personal Page: GitHub – SilvioM97

Location

• In-person: AIP Open Space
• Online: via Zoom

Abstract

Nearest Neighbors search addresses the problem of similarity search, a fundamental task in areas such as information retrieval, object recognition, and recommendation systems. However, scaling Nearest Neighbors search to large data collections poses a significant efficiency challenge.

Approximate Nearest Neighbors (ANN) search techniques address this by offering a valuable trade-off between speed and accuracy. In this talk, I will introduce kANNolo, a research-oriented ANN search library designed to accelerate and simplify innovation in ANN methodologies.

By presenting kANNolo’s architecture and core functionalities, I will highlight how this library can serve as a foundation for future advances in the field — from the deployment of ANN solutions on resource-constrained devices to the exploration of novel indexing algorithms and compression strategies.

Ultimately, this presentation aims to offer insights into the next generation of ANN research and development, underscoring both the practical and theoretical dimensions of similarity search at scale.