CacheDiff: Fast Random Sampling

TL;DR

CacheDiff is a novel sampling method that efficiently selects k random items from a known pool of N items with optimal time and space complexity, suitable for large datasets.

Contribution

The paper introduces CacheDiff, a new sampling algorithm with O(k) time and space complexity, improving efficiency over existing methods.

Findings

Achieves constant time per sample after preprocessing

Uses minimal memory proportional to k

Suitable for large-scale data sampling

Abstract

We present a sampling method called, CacheDiff, that has both time and space complexity of O(k) to randomly select k items from a pool of N items, in which N is known.