Abstract
The remarkable performance of convolutional neural networks (CNNs) is entangled with their huge number of uninterpretable parameters, which has become the bottleneck limiting the exploitation of their full potential. Towards network interpretation, previous endeavors mainly resort to the single filter analysis, which however ignores the relationship between filters. In this paper, we propose a novel architecture decoupling method to interpret the network from a perspective of investigating its calculation paths. More specifically, we introduce a novel architecture controlling module in each layer to encode the network architecture by a vector. By maximizing the mutual information between the vectors and input images, the module is trained to select specific filters to distill a unique calculation path for each input. Furthermore, to improve the interpretability and compactness of the decoupled network, the output of each layer is encoded to align the architecture encoding vector with the constraint of sparsity regularization. Unlike conventional pixel-level or filter-level network interpretation methods, we propose a path-level analysis to explore the relationship between the combination of filter and semantic concepts, which is more suitable to interpret the working rationale of the decoupled network. Extensive experiments show that the decoupled network achieves several applications, i.e., network interpretation, network acceleration, and adversarial samples detection.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Agakov, D.B.F.: The im algorithm: a variational approach to information maximization. In: NeurIPS (2004)
Bau, D., Zhou, B., Khosla, A., Oliva, A., Torralba, A.: Network dissection: quantifying interpretability of deep visual representations. In: CVPR (2017)
Bengio, Y., Léonard, N., Courville, A.: Estimating or propagating gradients through stochastic neurons for conditional computation. arXiv preprint arXiv:1308.3432 (2013)
Bolukbasi, T., Wang, J., Dekel, O., Saligrama, V.: Adaptive neural networks for efficient inference. In: ICML (2017)
Chen, R., Chen, H., Huang, G., Ren, J., Zhang, Q.: Explaining neural networks semantically and quantitatively. In: ICCV (2019)
Chen, Z., Li, Y., Bengio, S., Si, S.: You look twice: gaternet for dynamic filter selection in CNNS. In: CVPR (2019)
Dong, X., Huang, J., Yang, Y., Yan, S.: More is less: a more complicated network with less inference complexity. In: CVPR (2017)
Dosovitskiy, A., Brox, T.: Inverting visual representations with convolutional networks. In: CVPR (2016)
Figurnov, M., et al.: Spatially adaptive computation time for residual networks. In: CVPR (2017)
Fong, R., Vedaldi, A.: Net2vec: quantifying and explaining how concepts are encoded by filters in deep neural networks. In: CVPR (2018)
Gao, X., Zhao, Y., Dudziak, L., Mullins, R., Xu, C.Z.: Dynamic channel pruning: feature boosting and suppression. ICLR (2018)
Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. ICLR (2015)
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: CVPR (2016)
He, Y., Kang, G., Dong, X., Fu, Y., Yang, Y.: Soft filter pruning for accelerating deep convolutional neural networks. IJCAI (2018)
He, Y., Zhang, X., Sun, J.: Channel pruning for accelerating very deep neural networks. In: ICCV (2017)
Hu, J., Shen, L., Sun, G.: Squeeze-and-excitation networks. In: CVPR (2018)
Ioffe, S., Szegedy, C.: Batch normalization: accelerating deep network training by reducing internal covariate shift. Int. Conf. Mach. Learn. (2015)
Kaiser, Ł., Bengio, S.: Can active memory replace attention? In: NeurIPS (2016)
Kaiser, Ł., et al.: Fast decoding in sequence models using discrete latent variables. ICML (2018)
Koh, P.W., Liang, P.: Understanding black-box predictions via influence functions. ICML (2017)
Krizhevsky, A., Hinton, G.: Learning multiple layers of features from tiny images. Tech. rep, Citeseer (2009)
Lakkaraju, H., Kamar, E., Caruana, R., Horvitz, E.: Identifying unknown unknowns in the open world: representations and policies for guided exploration. In: AAAI (2017)
Li, H., Kadav, A., Durdanovic, I., Samet, H., Graf, H.P.: Pruning filters for efficient convnets. ICLR (2016)
Liu, L., Deng, J.: Dynamic deep neural networks: optimizing accuracy-efficiency trade-offs by selective execution. In: AAAI (2018)
Liu, Z., Li, J., Shen, Z., Huang, G., Yan, S., Zhang, C.: Learning efficient convolutional networks through network slimming. In: ICCV (2017)
Lundberg, S.M., Lee, S.I.: A unified approach to interpreting model predictions. In: NeurIPS (2017)
Maaten, L., Hinton, G., Visualizing data using t-SNE: Visualizing data using t-SNE. J. Mach. Learn. Res. 9, 2579–2605 (2008)
Morcos, A.S., Barrett, D.G., Rabinowitz, N.C., Botvinick, M.: On the importance of single directions for generalization. ICLR (2018)
Nair, V., Hinton, G.E.: Rectified linear units improve restricted boltzmann machines. In: International Conference on Machine Learning (2010)
Paszke, A., et al.: Automatic differentiation in pytorch. NeurIPS Workshop (2017)
Russakovsky, O., et al.: Imagenet large scale visual recognition challenge. IJCV 115(3), 211–252 (2015)
Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014)
Szegedy, C., et al.: Going deeper with convolutions. In: CVPR (2015)
Wang, J., Zhang, Z., Xie, C., Premachandran, V., Yuille, A.: Unsupervised learning of object semantic parts from internal states of cnns by population encoding. arXiv preprint arXiv:1511.06855 (2015)
Wang, X., Yu, F., Dou, Z.Y., Darrell, T., Gonzalez, J.E.: Skipnet: learning dynamic routing in convolutional networks. In: ECCV (2018)
Wang, Y., Su, H., Zhang, B., Hu, X.: Interpret neural networks by identifying critical data routing paths. In: CVPR (2018)
Yiyou, S., Sathya N., R., Vikas, S.: Adaptive activation thresholding: dynamic routing type behavior for interpretability in convolutional neural networks. In: ICCV (2019)
Yosinski, J., Clune, J., Bengio, Y., Lipson, H.: How transferable are features in deep neural networks? In: NeurIPS (2014)
Yosinski, J., Clune, J., Nguyen, A., Fuchs, T., Lipson, H.: Understanding neural networks through deep visualization. Int. Conf. Mach. Learn. Workshop (2015)
Zeiler, M.D., Fergus, R.: Visualizing and understanding convolutional networks. In: Fleet, D., Pajdla, T., Schiele, B., Tuytelaars, T. (eds.) ECCV 2014. LNCS, vol. 8689, pp. 818–833. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-10590-1_53
Zhang, Q., Nian Wu, Y., Zhu, S.C.: Interpretable convolutional neural networks. In: CVPR (2018)
Zhang, Q., Yang, Y., Wu, Y.N., Zhu, S.C.: Interpreting cnns via decision trees. In: CVPR (2019)
Zhuang, Z., et al.: Discrimination-aware channel pruning for deep neural networks. In: NeurIPS (2018)
Acknowledgements
This work is supported by the Nature Science Foundation of China (No. U1705262, No. 61772443, No. 61572410, No. 61802324 and No. 61702136), National Key R&D Program (No. 2017YFC0113000, and No. 2016Y FB1001503), Key R&D Program of Jiangxi Province (No. 20171ACH80022) and Natural Science Foundation of Guangdong Provice in China (No. 2019B1515120049).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
1 Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Li, Y. et al. (2020). Interpretable Neural Network Decoupling. In: Vedaldi, A., Bischof, H., Brox, T., Frahm, JM. (eds) Computer Vision – ECCV 2020. ECCV 2020. Lecture Notes in Computer Science(), vol 12360. Springer, Cham. https://doi.org/10.1007/978-3-030-58555-6_39
Download citation
DOI: https://doi.org/10.1007/978-3-030-58555-6_39
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-58554-9
Online ISBN: 978-3-030-58555-6
eBook Packages: Computer ScienceComputer Science (R0)