Realigned Softmax Warping for Deep Metric Learning
This work addresses the challenge of optimizing embedding spaces in deep metric learning, offering a novel approach that is incremental but shows strong performance improvements.
The paper tackles the problem of controlling separability and compactness forces in deep metric learning by proposing a new class of loss functions that use a warping function to manipulate these forces, achieving competitive, state-of-the-art results on various benchmarks.
Deep Metric Learning (DML) loss functions traditionally aim to control the forces of separability and compactness within an embedding space so that the same class data points are pulled together and different class ones are pushed apart. Within the context of DML, a softmax operation will typically normalize distances into a probability for optimization, thus coupling all the push/pull forces together. This paper proposes a potential new class of loss functions that operate within a euclidean domain and aim to take full advantage of the coupled forces governing embedding space formation under a softmax. These forces of compactness and separability can be boosted or mitigated within controlled locations at will by using a warping function. In this work, we provide a simple example of a warping function and use it to achieve competitive, state-of-the-art results on various metric learning benchmarks.