Symmetrical Synthesis for Deep Metric Learning
This work addresses a specific problem in deep metric learning for researchers and practitioners by offering a plug-and-play solution to enhance embedding quality without network modifications, though it is incremental in nature.
The paper tackles the issues of hyper-parameter tuning, optimization difficulty, and slow training in deep metric learning by proposing symmetrical synthesis, a hyper-parameter-free method for generating synthetic hard samples, which improves performance on clustering and image retrieval tasks over existing methods.
Deep metric learning aims to learn embeddings that contain semantic similarity information among data points. To learn better embeddings, methods to generate synthetic hard samples have been proposed. Existing methods of synthetic hard sample generation are adopting autoencoders or generative adversarial networks, but this leads to more hyper-parameters, harder optimization, and slower training speed. In this paper, we address these problems by proposing a novel method of synthetic hard sample generation called symmetrical synthesis. Given two original feature points from the same class, the proposed method firstly generates synthetic points with each other as an axis of symmetry. Secondly, it performs hard negative pair mining within the original and synthetic points to select a more informative negative pair for computing the metric learning loss. Our proposed method is hyper-parameter free and plug-and-play for existing metric learning losses without network modification. We demonstrate the superiority of our proposed method over existing methods for a variety of loss functions on clustering and image retrieval tasks. Our implementations is publicly available.