Abstract
Convolutional neural networks (CNNs) classify images by learning intermediate representations of the input throughout many layers. In recent work, latent representations of CNNs have been aligned with semantic concepts. However, for generating such alignments, the majority of existing methods predominantly rely on large amounts of labeled data, which is hard to acquire in practice. In this work, we address this limitation by presenting a framework for mapping hidden units from CNNs to semantic attributes of classes extracted from external commonsense knowledge repositories. We empirically demonstrate the effectiveness of our framework on copy-paste adversarial image classification and generalized zero-shot learning tasks.
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
Notes
- 1.
Alternatively, a KG \(\mathcal {G}=(\mathcal {V}, \{\mathcal {E}_p \subseteq \mathcal {V} \times \mathcal {V}\}_{p \in \mathcal {P}})\) can be viewed as a directed super-graph (i.e. a composition of directed graphs \(\mathcal {G}_p=(\mathcal {V}, \mathcal {E}_p), \forall p \in \mathcal {P}\), where the edges are labeled by the predicates p.
- 2.
We also write \(J(\mathcal {X},\mathcal {Y})\) for conciseness.
- 3.
A neuron can also be understood as an element of the vector of activation output for a given layer.
- 4.
We also experimented with the WebChild [27] KG, but the results for ConceptNet are more promising.
References
Agrawal, R., Imielinski, T., Swami, A.N.: Mining association rules between sets of items in large databases. In: SIGMOD 1993, pp. 207–216 (1993)
Bau, D., Zhou, B., Khosla, A., Oliva, A., Torralba, A.: Network dissection: quantifying interpretability of deep visual representations. In: CVPR, pp. 3319–3327 (2017)
Brunner, T., Diehl, F., Knoll, A.: Copy and paste: a simple but effective initialization method for black-box adversarial attacks. In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) (2019)
Chen, J., Geng, Y., Chen, Z., Horrocks, I., Pan, J.Z., Chen, H.: Knowledge-aware zero-shot learning: survey and perspective. In: IJCAI, pp. 4366–4373. ijcai.org (2021)
Cheng, X., Lu, J., Feng, J., Yuan, B., Zhou, J.: Scene recognition with objectness. Pattern Recogn. 74, 474–487 (2018)
Dalvi, F., Nortonsmith, A., Bau, A., et al.: Neurox: a toolkit for analyzing individual neurons in neural networks. In: AAAI 2019, pp. 9851–9852 (2019)
Deng, J., Dong, W., Socher, R., Li, L., et al.: Imagenet: a large-scale hierarchical image database. In: CVPR 2009, pp. 248–255 (2009)
Endres, D., Földiák, P.: Interpreting the neural code with formal concept analysis. In: NIPS, pp. 425–432 (2008)
Fong, R., Vedaldi, A.: Net2vec: quantifying and explaining how concepts are encoded by filters in deep neural networks. In: CVPR, pp. 8730–8738 (2018)
Geng, Y., Chen, J., Zhiquan Ye, e.: Explainable zero-shot learning via attentive graph convolutional network and KGs. SW 12 (2021)
Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. In: ICLR 2015 (2015)
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)
Horta, V.A.C., Mileo, A.: Towards explaining deep neural networks through graph analysis. In: DB and Expert Systems Applications, pp. 155–165 (2019)
Kampffmeyer, M., Chen, Y., Liang, X., Wang, H., Zhang, Y., Xing, E.P.: Rethinking knowledge graph propagation for zero-shot learning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2019)
Lécué, F.: On the role of knowledge graphs in explainable AI. Semant. Web 11(1), 41–51 (2020)
Mu, J., Andreas, J.: Compositional explanations of neurons. Adv. Neural Inf. Process. Syst. 33, 17153–17163 (2020)
Nayak, N.V., Bach, S.H.: Zero-shot learning with common sense knowledge graphs. CoRR abs/2006.10713 (2020)
Nguyen, A.M., Dosovitskiy, A., Jason Yosinski, e.: Synthesizing the preferred inputs for neurons in neural networks via deep generator networks. In: Neurips 2016, pp. 3387–3395 (2016)
Quattoni, A., Torralba, A.: Recognizing indoor scenes. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 413–420 (2009). https://doi.org/10.1109/CVPR.2009.5206537
Roy, A., Ghosal, D., Cambria, E., Majumder, N., Mihalcea, R., Poria, S.: Improving zero shot learning baselines with commonsense knowledge. CoRR abs/2012.06236 (2020)
Sarker, M.K., Xie, N., Doran, D., Raymer, M., Hitzler, P.: Explaining trained neural networks with semantic web technologies: first steps. In: NeSy (2017)
Selvaraju, R.R., et al.: Choose your neuron: incorporating domain knowledge through neuron-importance. In: ECCV (13), pp. 540–556 (2018)
Simonyan, K., Vedaldi, A., Zisserman, A.: Deep inside convolutional networks: visualising image classification models and saliency maps. In: ICLR 2014 (2014)
Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. In: ICLR 2015 (2015)
de Sousa Ribeiro, M., Leite, J.: Aligning artificial neural networks and ontologies towards explainable AI. In: AAAI 2021, pp. 4932–4940 (2021)
Speer, R., Chin, J., Havasi, C.: Conceptnet 5.5: an open multilingual graph of general knowledge. In: AAAI 2017, pp. 4444–4451 (2017)
Tandon, N., de Melo, G., Suchanek, F.M., Weikum, G.: Webchild: harvesting and organizing commonsense knowledge from the web. In: WSDM. ACM (2014)
Xian, Y., Lampert, C.H., Schiele, B., Akata, Z.: Zero-shot learning-a comprehensive evaluation. IEEE Trans. Pattern. Anal. Mach. Intell. 41(9), 2251–2265 (2019)
Zhou, B., Khosla, A., Lapedriza, À., Oliva, A., Torralba, A.: Object detectors emerge in deep scene cnns. In: ICLR 2015 (2015)
Acknowledgements
We would like to thank Dr. Volker Fischer from Bosch Center for AI for providing helpful feedback on initial versions of this work.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Ismaeil, Y., Stepanova, D., Tran, TK., Saranrittichai, P., Domokos, C., Blockeel, H. (2022). Towards Neural Network Interpretability Using Commonsense Knowledge Graphs. In: Sattler, U., et al. The Semantic Web – ISWC 2022. ISWC 2022. Lecture Notes in Computer Science, vol 13489. Springer, Cham. https://doi.org/10.1007/978-3-031-19433-7_5
Download citation
DOI: https://doi.org/10.1007/978-3-031-19433-7_5
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-19432-0
Online ISBN: 978-3-031-19433-7
eBook Packages: Computer ScienceComputer Science (R0)
