All Topics
All Topics
Technology
Technology
Design
Design
Programming
Programming
Science
Science
News
News
Gaming
Gaming
Entertainment
Entertainment
Business
Business
Finance
Finance
Sports
Sports
Health
Health
Food
Food
Travel
Travel
Art
Art
Music
Music
Books
Books
Education
Education
Politics
Politics
Personal
Personal
No algorithm. No AI slop. No ads. Just RSS. Pro-human. Indie writers. Real journalism. Open web. Chronological. Hand toasted.

Theoretical Constraints of Embedding-Based Retrieval: Orthogonality Limits in Vector Spaces

By

sonabinu

9mo ago· 1 min readzhInsight

Summary

This article discusses the theoretical limitations of embedding-based retrieval systems, focusing on the mathematical constraints of vector representations in high-dimensional spaces. It explains the concept of strict orthogonality in n-dimensional Euclidean space where vectors must have exact 90-degree angles (dot product of 0), limiting the maximum number of orthogonal vectors to n. The article then explores approximate orthogonality where vectors can have angles close to 90 degrees (e.g., 89-91 degrees), which allows for more vectors while maintaining useful mathematical properties for retrieval systems.

Key quotes

· 4 pulled
在 n 维欧几里得空间 R^n 中,一个向量组如果两两严格正交(夹角精确为90°,或点积为0),那么这个向量组的大小(基数)最多为 n
如果我们放宽这个条件,只要求向量之间的夹角"接近"90°,例如在你说的89°到91°之间
这构成了该空间的一组正交基
要求任意两个不同的单位向量 v...
Snippet from the RSS feed
View 1 comments: 严格正交 (Orthogonal):在 n 维欧几里得空间 R^n 中,一个向量组如果两两严格正交(夹角精确为90°,或点积为0),那么这个向量组的大小(基数)最多为 n。这构成了该空间的一组正交基。近似正交 (Almost/Nearly Orthogonal):如果我们放宽这个条件,只要求向量之间的夹角“接近”90°,例如在你说的89°到91°之间,或者更形式化地,要求任意两个不同的单位向量 v...

You might also wanna read