本篇論文針對圖像檢索(Image retrieval )的任務上提出了一個新的損失函數。此方法基於Proxy_NCA以及Proxy_Anchor的方法上加上了多個代表點的方法，來提升樣本的豐富性。使得Batch size減少的情況下也能達到跟原來較大的batch size一樣的效果。並且使用SoftMax函數對類內代表點做加權。使得重要的代表點能得到更多的學習資源。除此損失函數地改良之外，也對現有的ResNet50進行了修改，只使用RestNet50的前三層做為特徵擷取，取消了ResNet50第三層的下採樣。並且加入了Attention機制取代原本ResNet50的第四層。Attention使用了SoftPlus函數對特徵圖的特徵做加權。使得重要的特徵能更明顯，不重要的特徵減少關注度。 相較於傳統Attention使用SoftMax函數能得到更好的效果。不管是新提出的損失函數，或是改良過後的ResNet50都相較於原始方法Recall@1都有很大的提升。

In this paper, we propose a new loss function for Image Retrieval task. The new loss function makes an improvement based on Proxy-NCA and Proxy-Anchor Loss by adopting multiple proxies, to promote positive sample variety. Its shows better performance than Proxy-Anchor Loss even in the small batch size. Besides, we weighted intra-class proxy by SoftMax function to make important samples receive a higher gradient while training. In addition, we make some changes on ResNet50 by only using the first three-layer and adding a new attention module by using SoftPlus function to replace SoftMax. Finally, we obtain well results on recall@1 via our new method.

[1] S. Na, L. Xumin, G. Yong. Research on k-means Clustering Algorithm: An Improved k-means Clustering Algorithm. Third International Symposium on Intelligent Information Technology and Security Informatics, 2010, pp. 63-67, doi: 10.1109/IITSI.2010.74.
[2] M. A. Hearst, S. T. Dumais, E. Osuna, J. Platt and B. Scholkopf, Support vector machines. IEEE Intelligent Systems and their Applications, 1998, pp. 18-28, doi: 10.1109/5254.708428.
[3] H, Jegou. M, Douze. and C , Schmid. Product quantization for nearest neighbor search. IEEE Trans. Pattern Anal. Mach. Intell.
[4] F. Schroff, D. Kalenichenko and J. Philbin. FaceNet: A unified embedding for face recognition and clustering.Conference on Computer Vision and Pattern Recognition (CVPR), 2015, pp. 815-823, doi: 10.1109
[5] X, Wang. X, Han. W, Huang. D, Dong. and M,R Scott. Multi-similarity loss with general pair weighting for deep metric learning. Conference on Computer Vision and Pattern Recognition (CVPR), 2019.
[6] Y. Movshovitz-Attias, A. Toshev, T. Leung, S. Ioffe and S. Singh. No Fuss Distance Metric Learning Using Proxies. International Conference on Computer Vision (ICCV), Venice, Italy, 2017 pp. 360-368.doi: 10.1109
[7] Q, Qian. L, Shang. B, Sun. J, Hu.T, Tacoma. H, Li, and R, Jin. Softtriple loss: Deep metric learning without triplet sampling. International Conference on Computer Vision (ICCV), 2019.
[8] K,Sungyeon. K, Dongwon. C, Minsu. K, Suha. Proxy Anchor Loss for Deep Metric Learning. https://doi.org/10.48550/arXiv.2003.13911
[9] Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal covariate shift. International Conference on Machine Learning (ICML), 2015.
[10] H, Kaiming. Z, Xiangyu. R, Shaoqing. and S, Jian. Deep residual learning for image recognition. Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
[11]G,Thomas. Dietterich Oregon State University Corvallis Oregon USA tgdcsorstedu WWW home page http://www.cs.orst.edu/tgd
[12] B, Lucas and H, Alexander contribute https: //doi.org/10.48550/arXiv 1703.07737
[13]V, Ashish. S, Noam. P, Niki. J, Parmar. L, Uszkoreit. A, Jones. N. Gomez, K, Lukasz and P, Illia. https://doi.org/10.48550/arXiv.1706.03762
[14] J, Krause., M, Stark., J, Deng and L, Fei-Fei. 3d object representations for fine-grained categorization. Workshop on 3D Representation and Recognition, 2013.
[15] C, Wah. S, Branson. P, Welinder. P, Perona. and S, Belongie. The Caltech-UCSD Birds-200-2011 Dataset. Technical Report CNS-TR-2011-001, California Institute of Technology, 2011.
[16] D, Jia. D, Wei. S, Richard. L, Li-Jia. L, Kai and L, Fei-Fei. ImageNet: a large-scale hierarchical image database. Conference on Computer Vision and Pattern Recognition (CVPR), 2009.
[17] L, Ilya. and H, Frank . Decoupled weight decay regularization. International Conference on Learning Representations (ICLR), 2019.
[18] A.F, McDaid. D, Greene., and N,J Hurley. Normalized mutual information to evaluate overlapping community finding algorithms. CoRR, abs/1110.2515, 2011.
[19] O, Song. Hyun. Deep metric learning via lifted structured feature embedding. Conference on Computer Vision and Pattern Recognition(CVPR), 2016.
[20] S, Hyun Oh. J, Stefanie. R, Vivek. and M, Kevin. Deep metric learning via facility location.Conference on Computer Vision and Pattern Recognition (CVPR), 2017.
[21] K, Sohn. Improved deep metric learning with multi-class n-pair loss objective. Neural Information Processing Systems (NIPS), 2016.
[22] C. Szegedy et al.Going deeper with convolutions.Conference on Computer Vision and Pattern Recognition (CVPR), 2015, pp. 1-9, doi: 10.1109.
[23] J, Wang. F, Zhou. S, Wen. X, Liu. and Y, Lin. Deep metric learning with angular loss. International Conference on Computer Vision (ICCV), 2017.
[24] M,T Law. R, Urtasun. and R,S Zemel. Deep spectral clustering learning. International Conference on Machine Learning (ICML), 2017.
[25] G, Weifeng. H, Weilin. D, Dengke, and S, Matthew. Deep metric learning with hierarchical triplet loss. European Conference on Computer Vision (ECCV), 2018.
[26] A, Zhai. and H, Wu. Classification is a strong baseline for deep metric learning. British Machine Vision Conference (BMVC), 2019.
[27] P, Jacob. D, Picard. A, Histace. and E, Klein. Metric learning with HORDE: high-order regularizer for deep embeddings. International Conference on Computer Vision (ICCV), 2019.
[28] H, Xuan. A, Stylianou. and R, Pless. Improved embeddings with easy positive triplet mining. Winter Conference on Applications of Computer Vision (WACV), 2020.
[29] I, Elezi. S, Vascon. A, Torchinovich. M, Pelillo. and L, LealTaixe. The group loss for deep metric learning. European Conference in Computer Vision (ECCV), 2020.
[30] E, Teh. T, DeVries., and G,W Taylor. Proxynca++: Revisiting and revitalizing proxy neighborhood component analysis. European Conference on Computer Vision (ECCV), 2020.
[31] Y, Zhu., M, Yang. C, Deng. and W, Liu. Fewer is more: A deep graph metric learning perspective using fewer proxies. Conference and Workshop on Neural Information Processing Systems (NeurIPS), 2020.