Year: 2024
Type: preprint
Abstract: Traditional database management systems need help efficiently represent and querying the complex, high-dimensional data prevalent in modern applications. Vector databases offer a solution by storing data as numerical vectors within a multi-dimensional space. This enables similarity-based search and analysis, such as image retrieval, recommendation engine generation, and natural language processing. This paper introduces Quantixar, a vector database project designed for efficiency in high-dimensional settings. Quantixar tackles the challenge of managing high-dimensional data by strategically combining advanced indexing and quantization techniques. It employs HNSW indexing for accelerated ANN search. Additionally, Quantixar incorporates binary and product quantization to compress high-dimensional vectors, reducing storage requirements and computational costs during search. The paper delves into Quantixar's architecture, specific implementation, and experimental methodology.
Source: arXiv (Cornell University)
Authors: Yadav, Gulshan, Yadav, RahulKumar, Viramgama, Mansi, Viramgama, Mayank, Mohite, Apeksha
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4393027404
