Diversified Mutual Learning for Deep Metric Learning
This work addresses the challenge of inductive transfer learning in deep metric learning, particularly when large-scale data is scarce, by improving model performance and ensemble results.
The paper tackles the problem of improving generalization in deep metric learning by proposing Diversified Mutual Metric Learning, which enhances embedding models through diversified mutual learning, achieving state-of-the-art Recall@1 scores of 69.9 on CUB-200-2011 and 89.1 on CARS-196.
Mutual learning is an ensemble training strategy to improve generalization by transferring individual knowledge to each other while simultaneously training multiple models. In this work, we propose an effective mutual learning method for deep metric learning, called Diversified Mutual Metric Learning, which enhances embedding models with diversified mutual learning. We transfer relational knowledge for deep metric learning by leveraging three kinds of diversities in mutual learning: (1) model diversity from different initializations of models, (2) temporal diversity from different frequencies of parameter update, and (3) view diversity from different augmentations of inputs. Our method is particularly adequate for inductive transfer learning at the lack of large-scale data, where the embedding model is initialized with a pretrained model and then fine-tuned on a target dataset. Extensive experiments show that our method significantly improves individual models as well as their ensemble. Finally, the proposed method with a conventional triplet loss achieves the state-of-the-art performance of Recall@1 on standard datasets: 69.9 on CUB-200-2011 and 89.1 on CARS-196.