Pairwise Similarity Learning Is Simple

Abstract

In this paper, we focus on a general yet important learning problem, pairwise similarity learning (PSL). PSL subsumes a wide range of important applications, such as open-set face recognition, speaker verification, image retrieval and person re-identification. The goal of PSL is to learn a pairwise similarity function assigning a higher similarity score to positive pairs (i.e., a pair of samples with the same label) than to negative pairs (i.e., a pair of samples with different label). We start by identifying a key desideratum for PSL, and then discuss how existing methods can achieve this desideratum. We then propose a surprisingly simple proxy-free method, called SimPLE, which requires neither feature/proxy normalization nor angular margin and yet is able to generalize well in open-set recognition. We apply the proposed method to three challenging PSL tasks: open-set face recognition, image retrieval and speaker verification. Comprehensive experimental results on large-scale benchmar

Stats

• citations9
• S2 citations—
• github stars0
• HF likes0
• heat score7.50
• arxiv keywen2023pairwise

Related papers

• Sparse Online Relative Similarity Learning (2021)2.26
• Compare More Nuanced:pairwise Alignment Bilinear Network For Few-shot Fine-grained Learning (2019)11.39
• Large Margin Learning In Set To Set Similarity Comparison For Person Re-identification (2017)12.10
• Effectively Leveraging Attributes For Visual Similarity (2021)3.66
• Super-sparse Learning In Similarity Spaces (2017)5.24
• Learnt Quasi-transitive Similarity For Retrieval From Large Collections Of Faces (2016)5.24
• Multi-level Similarity Learning For Low-shot Recognition (2019)0.00
• Attributable Visual Similarity Learning (2022)11.68