Distance-Ratio-Based Formulation for Metric Learning
This work addresses metric learning for few-shot classification, offering incremental improvements in training efficiency and stability.
The authors tackled the problem of metric learning by proposing a distance-ratio-based formulation that is scale-invariant and outputs optimal classification confidence scores, resulting in faster and more stable learning with improved or comparable generalization on CUB and mini-ImageNet datasets.
In metric learning, the goal is to learn an embedding so that data points with the same class are close to each other and data points with different classes are far apart. We propose a distance-ratio-based (DR) formulation for metric learning. Like softmax-based formulation for metric learning, it models $p(y=c|x')$, which is a probability that a query point $x'$ belongs to a class $c$. The DR formulation has two useful properties. First, the corresponding loss is not affected by scale changes of an embedding. Second, it outputs the optimal (maximum or minimum) classification confidence scores on representing points for classes. To demonstrate the effectiveness of our formulation, we conduct few-shot classification experiments using softmax-based and DR formulations on CUB and mini-ImageNet datasets. The results show that DR formulation generally enables faster and more stable metric learning than the softmax-based formulation. As a result, using DR formulation achieves improved or comparable generalization performances.