Saved in:
Bibliographic Details
Main Authors: Deuser, Fabian, Hausenblas, Philipp, Schieber, Hannah, Roth, Daniel, Werner, Martin, Oswald, Norbert
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.17844
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913855651708928
author Deuser, Fabian
Hausenblas, Philipp
Schieber, Hannah
Roth, Daniel
Werner, Martin
Oswald, Norbert
author_facet Deuser, Fabian
Hausenblas, Philipp
Schieber, Hannah
Roth, Daniel
Werner, Martin
Oswald, Norbert
contents Contrastive learning is a representational learning paradigm in which a neural network maps data elements to feature vectors. It improves the feature space by forming lots with an anchor and examples that are either positive or negative based on class similarity. Hard negative examples, which are close to the anchor in the feature space but from a different class, improve learning performance. Finding such examples of high quality efficiently in large, high-dimensional datasets is computationally challenging. In this paper, we propose a GPU-friendly Locality-Sensitive Hashing (LSH) scheme that quantizes real-valued feature vectors into binary representations for approximate nearest neighbor search. We investigate its theoretical properties and evaluate it on several datasets from textual and visual domain. Our approach achieves comparable or better performance while requiring significantly less computation than existing hard negative mining strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2505_17844
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Locality-Sensitive Hashing for Efficient Hard Negative Sampling in Contrastive Learning
Deuser, Fabian
Hausenblas, Philipp
Schieber, Hannah
Roth, Daniel
Werner, Martin
Oswald, Norbert
Computer Vision and Pattern Recognition
Contrastive learning is a representational learning paradigm in which a neural network maps data elements to feature vectors. It improves the feature space by forming lots with an anchor and examples that are either positive or negative based on class similarity. Hard negative examples, which are close to the anchor in the feature space but from a different class, improve learning performance. Finding such examples of high quality efficiently in large, high-dimensional datasets is computationally challenging. In this paper, we propose a GPU-friendly Locality-Sensitive Hashing (LSH) scheme that quantizes real-valued feature vectors into binary representations for approximate nearest neighbor search. We investigate its theoretical properties and evaluate it on several datasets from textual and visual domain. Our approach achieves comparable or better performance while requiring significantly less computation than existing hard negative mining strategies.
title Locality-Sensitive Hashing for Efficient Hard Negative Sampling in Contrastive Learning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.17844