Pairwise Similarity Learning is SimPLE
This addresses a general learning problem with broad applications in computer vision and speech, but it appears incremental as it builds on existing PSL methods.
The paper tackles pairwise similarity learning (PSL) for applications like face recognition and image retrieval by proposing SimPLE, a proxy-free method that avoids normalization and angular margins, achieving state-of-the-art results on large-scale benchmarks.
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 benchmarks show that our method performs significantly better than current state-of-the-art methods.