Unsupervised Deep Metric Learning via Auxiliary Rotation Loss
This addresses the need for labeled data in metric learning for domains like image retrieval, though it is incremental as it builds on existing clustering and self-supervision techniques.
The paper tackles the problem of requiring labeled data for deep metric learning by proposing an unsupervised method that generates pseudo-labels from clustering and regularizes training with a self-supervised rotation prediction task, resulting in outperforming state-of-the-art approaches by a large margin on standard benchmarks.
Deep metric learning is an important area due to its applicability to many domains such as image retrieval and person re-identification. The main drawback of such models is the necessity for labeled data. In this work, we propose to generate pseudo-labels for deep metric learning directly from clustering assignment and we introduce unsupervised deep metric learning (UDML) regularized by a self-supervision (SS) task. In particular, we propose to regularize the training process by predicting image rotations. Our method (UDML-SS) jointly learns discriminative embeddings, unsupervised clustering assignments of the embeddings, as well as a self-supervised pretext task. UDML-SS iteratively cluster embeddings using traditional clustering algorithm (e.g., k-means), and sampling training pairs based on the cluster assignment for metric learning, while optimizing self-supervised pretext task in a multi-task fashion. The role of self-supervision is to stabilize the training process and encourages the model to learn meaningful feature representations that are not distorted due to unreliable clustering assignments. The proposed method performs well on standard benchmarks for metric learning, where it outperforms current state-of-the-art approaches by a large margin and it also shows competitive performance with various metric learning loss functions.