Abstract
In model compression, knowledge distillation is most used technique that uses a large teacher network to transfer knowledge to a small student network to improve student network performance. However, most knowledge distillation algorithms only focus on exploring informative knowledge for transferring but ignore the consistency between the teacher network and the student network. In this paper, we propose a new knowledge distillation framework (SNKD) to calculate the consistency of the teacher network and the student network based on the siamese networks. The teacher network and student network features are input into the siamese networks to calculate the discrepancies between them based on the contrastive learning loss. Through minimizing the contrastive learning loss, the student network is promoted to consistent with the teacher network and obtain a ability close to the teacher. We have verify the efficiency of the SNKD by experiment on popular datasets. All SNKD trained student network models have reached ability similar or even better than teacher networks.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bertinetto, L., Valmadre, J., Henriques, J.F., Vedaldi, A., Torr, P.H.S.: Fully-convolutional siamese networks for object tracking. In: Hua, G., Jégou, H. (eds.) ECCV 2016. LNCS, vol. 9914, pp. 850–865. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-48881-3_56
Bromley, J., Guyon, I., LeCun, Y., Säckinger, E., Shah, R.: Signature verification using a siamese time delay neural network. Adv. Neural Inf. Process. Syst. 6, 737–744 (1993)
Coates, A., Ng, A., Lee, H.: An analysis of single-layer networks in unsupervised feature learning. In: Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, pp. 215–223. JMLR Workshop and Conference Proceedings (2011)
He, K., Zhang, X., Ren, S., Jian, S.: Deep residual learning for image recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016)
Heo, B., Lee, M., Yun, S., Choi, J.Y.: Knowledge transfer via distillation of activation boundaries formed by hidden neurons. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 33, pp. 3779–3787 (2019)
Hinton, G., Vinyals, O., Dean, J.: Distilling the knowledge in a neural network. Comput. Sci. 14(7), 38–39 (2015)
Hoffer, E., Ailon, N.: Deep metric learning using triplet network. In: Feragen, A., Pelillo, M., Loog, M. (eds.) SIMBAD 2015. LNCS, vol. 9370, pp. 84–92. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-24261-3_7
Krizhevsky, A., Hinton, G., et al.: Learning multiple layers of features from tiny images (2009)
Nair, V., Hinton, G.E.: Rectified linear units improve restricted boltzmann machines. In: ICML (2010)
Netzer, Y., Wang, T., Coates, A., Bissacco, A., Wu, B., Ng, A.Y.: Reading digits in natural images with unsupervised feature learning (2011)
Park, W., Kim, D., Lu, Y., Cho, M.: Relational knowledge distillation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 3967–3976 (2019)
Peng, B., et al.: Correlation congruence for knowledge distillation (2019)
Romero, A., Ballas, N., Kahou, S.E., Chassang, A., Gatta, C., Bengio, Y.: Fitnets: hints for thin deep nets. Comput. Sci. (2014)
Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014)
Tung, F., Mori, G.: Similarity-preserving knowledge distillation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 1365–1374 (2019)
Wang, K., Gao, X., Zhao, Y., Li, X., Dou, D., Xu, C.Z.: Pay attention to features, transfer learn faster CNNs. In: International Conference on Learning Representations (2019)
Xu, Z., Hsu, Y.C., Huang, J.: Training shallow and thin networks for acceleration via knowledge distillation with conditional adversarial networks (2017)
Zagoruyko, S., Komodakis, N.: Paying more attention to attention: Improving the performance of convolutional neural networks via attention transfer. arXiv preprint arXiv:1612.03928 (2016)
Zagoruyko, S., Komodakis, N.: Wide residual networks. arXiv preprint arXiv:1605.07146 (2016)
Acknowledgement
This work was supported by the Mianyang Science and Technology Program 2020YFZJ016, SWUST Doctoral Foundation under Grant 19zx7102, 21zx7114, Sichuan Science and Technology Program under Grant 2020YFS0307.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Tang, J., Yang, X., Cheng, X., Jiang, N., Yu, W., Zhang, P. (2021). Consistent Knowledge Distillation Based on Siamese Networks. In: Mantoro, T., Lee, M., Ayu, M.A., Wong, K.W., Hidayanto, A.N. (eds) Neural Information Processing. ICONIP 2021. Communications in Computer and Information Science, vol 1516. Springer, Cham. https://doi.org/10.1007/978-3-030-92307-5_38
Download citation
DOI: https://doi.org/10.1007/978-3-030-92307-5_38
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-92306-8
Online ISBN: 978-3-030-92307-5
eBook Packages: Computer ScienceComputer Science (R0)