Generalizable Embeddings with Cross-batch Metric Learning
This work addresses a domain-specific problem in deep metric learning for improving generalization to unseen classes, but it appears incremental as it builds on existing GAP methods.
The paper tackled the problem of learning generalizable embeddings for unseen classes in deep metric learning by reformulating global average pooling as a convex combination of learnable prototypes and regularizing with cross-batch learning. The result was validation on 4 benchmarks, showing competitive performance without specifying concrete numbers.
Global average pooling (GAP) is a popular component in deep metric learning (DML) for aggregating features. Its effectiveness is often attributed to treating each feature vector as a distinct semantic entity and GAP as a combination of them. Albeit substantiated, such an explanation's algorithmic implications to learn generalizable entities to represent unseen classes, a crucial DML goal, remain unclear. To address this, we formulate GAP as a convex combination of learnable prototypes. We then show that the prototype learning can be expressed as a recursive process fitting a linear predictor to a batch of samples. Building on that perspective, we consider two batches of disjoint classes at each iteration and regularize the learning by expressing the samples of a batch with the prototypes that are fitted to the other batch. We validate our approach on 4 popular DML benchmarks.