《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (12): 3639-3650.DOI: 10.11772/j.issn.1001-9081.2021091649
• 人工智能 • 下一篇
收稿日期:
2021-09-22
修回日期:
2022-01-15
接受日期:
2022-01-20
发布日期:
2022-12-21
出版日期:
2022-12-10
通讯作者:
刘军民
作者简介:
李凌敏(1998—),女,山西晋城人,硕士研究生,主要研究方向:神经网络可解释性基金资助:
Lingmin LI, Mengran HOU, Kun CHEN, Junmin LIU()
Received:
2021-09-22
Revised:
2022-01-15
Accepted:
2022-01-20
Online:
2022-12-21
Published:
2022-12-10
Contact:
Junmin LIU
About author:
LI Lingmin, born in 1998, M. S. candidate. Her research interests include interpretability of neural networks.Supported by:
摘要:
近年来,深度学习在很多领域得到广泛应用;然而,由于深度神经网络模型的高度非线性操作,导致其可解释性较差,并常常被称为“黑箱”模型,无法应用于一些对性能要求较高的关键领域;因此,对深度学习的可解释性开展研究是很有必要的。首先,简单介绍了深度学习;然后,围绕深度学习的可解释性,从隐层可视化、类激活映射(CAM)、敏感性分析、频率原理、鲁棒性扰动测试、信息论、可解释模块和优化方法这8个方面对现有研究工作进行分析;同时,展示了深度学习在网络安全、推荐系统、医疗和社交网络领域的应用;最后,讨论了深度学习可解释性研究存在的问题及未来的发展方向。
中图分类号:
李凌敏, 侯梦然, 陈琨, 刘军民. 深度学习的可解释性研究综述[J]. 计算机应用, 2022, 42(12): 3639-3650.
Lingmin LI, Mengran HOU, Kun CHEN, Junmin LIU. Survey on interpretability research of deep learning[J]. Journal of Computer Applications, 2022, 42(12): 3639-3650.
激活函数 | 基本形式 |
---|---|
Sigmoid | |
tanh | |
ReLU | |
Leaky ReLU | |
RReLU | |
Noisy ReLU |
表1 激活函数的形式
Tab. 1 Forms of activation functions
激活函数 | 基本形式 |
---|---|
Sigmoid | |
tanh | |
ReLU | |
Leaky ReLU | |
RReLU | |
Noisy ReLU |
损失函数 | 基本形式 |
---|---|
CrossEntropy | |
MSE损失函数 | |
log损失函数 | |
Adaboost | |
Hinge损失函数 |
表2 损失函数的基本形式
Tab. 2 Basic forms of loss functions
损失函数 | 基本形式 |
---|---|
CrossEntropy | |
MSE损失函数 | |
log损失函数 | |
Adaboost | |
Hinge损失函数 |
方法 | 优点 | 缺点 |
---|---|---|
CAM[ | 支持任意输入大小,参数少,鲁棒性强 | 需要修改原模型结构,重新训练 |
Grad-CAM[ | 可以可视化任意结构的卷积神经网络 | 缺乏突出显示细微细节的能力 |
Grad-CAM++[ | 更适用于同类多目标情况 | 视觉上不够干净,标记了大量背景信息 |
Score-CAM[ | 摆脱了梯度依赖,背景中噪声减小 | — |
Ablation-CAM[ | 直接对特征图进行掩码操作,定位更准确 | 需要遍历特征图,计算耗时长 |
表3 基于类激活映射的5种方法优缺点对比
Tab.3 Advantages and disadvantages comparison among five methods based on class activation mapping
方法 | 优点 | 缺点 |
---|---|---|
CAM[ | 支持任意输入大小,参数少,鲁棒性强 | 需要修改原模型结构,重新训练 |
Grad-CAM[ | 可以可视化任意结构的卷积神经网络 | 缺乏突出显示细微细节的能力 |
Grad-CAM++[ | 更适用于同类多目标情况 | 视觉上不够干净,标记了大量背景信息 |
Score-CAM[ | 摆脱了梯度依赖,背景中噪声减小 | — |
Ablation-CAM[ | 直接对特征图进行掩码操作,定位更准确 | 需要遍历特征图,计算耗时长 |
方法 | 隐层激活函数 | 输入变量对输出变量的敏感性系数 | |
---|---|---|---|
文献[ | |||
文献[ | |||
文献[ |
表4 基于偏导的敏感性分析方法
Tab. 4 Sensitivity analysis methods based on partial derivatives
方法 | 隐层激活函数 | 输入变量对输出变量的敏感性系数 | |
---|---|---|---|
文献[ | |||
文献[ | |||
文献[ |
类别 | 方法 | 特点 |
---|---|---|
被动解释 | 基于隐层可视化 | 运用可视化方法生成人类能理解的图像,解释隐层的含义 |
基于类激活映射 | 对线型图线性加权获得类激活图,解释个体的分类决策 | |
基于敏感性分析 | 对输入变量施加扰动,评估特征的重要性 | |
基于鲁棒性扰动测试 | 解释精心设计过的新输入对模型预测的影响程度 | |
基于频率原理 | 研究频率信号的规律,解释神经网络训练过程中的偏好 | |
主动解释 | 基于可解释模块 | 额外引入可解释的网络模块,修改原有网络结构 |
基于优化方法 | 向损失函数中添加正则化项,利用相关的优化方法进行解释 | |
补充解释 | 基于信息论 | 将信息论领域的相关概念或术语整合到神经网络中,获得更多信息的解释 |
结合方法 | 文献[ 通过衡量神经元对良性和对抗例子的行为变化强度,从神经元敏感性的角度对深层模型的鲁棒性进行了解释; 文献[ |
表5 各类方法特点对比及结合方法描述
Tab. 5 Comparison of characteristics of various methods and description of combined methods
类别 | 方法 | 特点 |
---|---|---|
被动解释 | 基于隐层可视化 | 运用可视化方法生成人类能理解的图像,解释隐层的含义 |
基于类激活映射 | 对线型图线性加权获得类激活图,解释个体的分类决策 | |
基于敏感性分析 | 对输入变量施加扰动,评估特征的重要性 | |
基于鲁棒性扰动测试 | 解释精心设计过的新输入对模型预测的影响程度 | |
基于频率原理 | 研究频率信号的规律,解释神经网络训练过程中的偏好 | |
主动解释 | 基于可解释模块 | 额外引入可解释的网络模块,修改原有网络结构 |
基于优化方法 | 向损失函数中添加正则化项,利用相关的优化方法进行解释 | |
补充解释 | 基于信息论 | 将信息论领域的相关概念或术语整合到神经网络中,获得更多信息的解释 |
结合方法 | 文献[ 通过衡量神经元对良性和对抗例子的行为变化强度,从神经元敏感性的角度对深层模型的鲁棒性进行了解释; 文献[ |
数据集 | DT | RobDT | LCPA | BBM-RS |
---|---|---|---|---|
adult | 414.20 | 287.90 | 14.90 | 6.00 |
bank | 30.70 | 26.80 | 8.90 | 8.00 |
bank2 | 30.00 | 30.70 | 13.80 | 4.50 |
breastcancer | 15.20 | 7.40 | 6.00 | 11.00 |
Careval | 59.30 | 28.20 | 10.10 | 8.70 |
Compasbin | 67.80 | 33.70 | 5.40 | 7.60 |
diabetes | 31.20 | 27.90 | 6.00 | 2.10 |
ficobin | 30.60 | 59.60 | 6.40 | 11.80 |
heart | 20.30 | 13.60 | 11.90 | 9.50 |
ionosphere | 11.30 | 8.60 | 17.90 | 6.80 |
表6 不同算法的IC指标的部分实验结果
Tab. 6 Some experimental results of IC index among different algorithms
数据集 | DT | RobDT | LCPA | BBM-RS |
---|---|---|---|---|
adult | 414.20 | 287.90 | 14.90 | 6.00 |
bank | 30.70 | 26.80 | 8.90 | 8.00 |
bank2 | 30.00 | 30.70 | 13.80 | 4.50 |
breastcancer | 15.20 | 7.40 | 6.00 | 11.00 |
Careval | 59.30 | 28.20 | 10.10 | 8.70 |
Compasbin | 67.80 | 33.70 | 5.40 | 7.60 |
diabetes | 31.20 | 27.90 | 6.00 | 2.10 |
ficobin | 30.60 | 59.60 | 6.40 | 11.80 |
heart | 20.30 | 13.60 | 11.90 | 9.50 |
ionosphere | 11.30 | 8.60 | 17.90 | 6.80 |
1 | ZEILER M D, FERGUS R. Visualizing and understanding convolutional networks[C]// Proceedings of the 2014 European Conference on Computer Vision, LNCS 8689. Cham: Springer, 2014: 818-833. |
2 | 陈珂锐,孟小峰. 机器学习的可解释性[J]. 计算机研究与发展, 2020, 57(9): 1971-1986. 10.7544/issn1000-1239.2020.20190456 |
CHEN K R, MENG X F. Interpretation and understanding in machine learning[J]. Journal of Computer Research and Development, 2020, 57(9): 1971-1986. 10.7544/issn1000-1239.2020.20190456 | |
3 | 化盈盈,张岱墀,葛仕明. 深度学习模型可解释性的研究进展[J]. 信息安全学报, 2020, 5(3): 1-12. 10.19363/J.cnki.cn10-1380/tn.2020.05.01 |
HUA Y Y, ZHANG D C, GE S M. Research progress in the interpretability of deep learning models[J]. Journal of Cyber Security, 2020, 5(3): 1-12. 10.19363/J.cnki.cn10-1380/tn.2020.05.01 | |
4 | 曾春艳,严康,王志锋,等. 深度学习模型可解释性研究综述[J]. 计算机工程与应用, 2021, 57(8): 1-9. 10.3778/j.issn.1002-8331.2012-0357 |
ZENG C Y, YAN K, WANG Z F, et al. Survey of interpretability research on deep learning models[J]. Computer Engineering and Applications, 2021, 57(8): 1-9. 10.3778/j.issn.1002-8331.2012-0357 | |
5 | GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]// Proceedings of the 27th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2014: 2672-2680. |
6 | KIM B, DOSHI-VELEZ F. Interpretable machine learning: the fuss, the concrete and the questions [C]// Proceedings of the 32nd International Conference on Machine Learning. New York: JMLR.org, 2017: 1-13. |
7 | DOSHI-VELEZ F, KIM B. Towards a rigorous science of interpretable machine learning[EB/OL]. (2017-03-02) [2021-08-10].. 10.1007/978-3-319-98131-4_1 |
8 | MILLER T. Explanation in artificial intelligence: insights from the social sciences[J]. Artificial Intelligence, 2019, 267:1-38. 10.1016/j.artint.2018.07.007 |
9 | MOLNAR C. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable[M]. Morrisville: Lulu Press, 2020: 53-70. |
10 | BAU D, ZHOU B L, KHOSLA A, et al. Network dissection: quantifying interpretability of deep visual representations[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 3319-3327. 10.1109/cvpr.2017.354 |
11 | ZHANG Q S, CAO R M, SHI F, et al. Interpreting CNN knowledge via an explanatory graph[C]// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2018: 4454-4463. 10.1609/aaai.v32i1.11819 |
12 | ZHANG Q S, WANG X, CAO R M, et al. Extraction of an explanatory graph to interpret a CNN[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(11): 3863-3877. 10.1109/tpami.2020.2992207 |
13 | AUBRY M, RUSSELL B C. Understanding deep features with computer-generated imagery[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 2875-2883. 10.1109/iccv.2015.329 |
14 | RAUBER P E, FADEL S G, FALCÃO A X, et al. Visualizing the hidden activity of artificial neural networks[J]. IEEE Transactions on Visualization and Computer Graphics, 2017, 23(1): 101-110. 10.1109/tvcg.2016.2598838 |
15 | ZHOU B L, KHOSLA A, LAPEDRIZA A, et al. Learning deep features for discriminative localization[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 2921-2929. 10.1109/cvpr.2016.319 |
16 | SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017: 618-626. 10.1109/iccv.2017.74 |
17 | CHATTOPADHYAY A, SARKAR A, HOWLADER P, et al. Grad-CAM++: generalized gradient-based visual explanations for deep convolutional networks [C]// Proceedings of the 2018 IEEE Winter Conference on Applications of Computer Vision. Piscataway: IEEE, 2018: 839-847. 10.1109/wacv.2018.00097 |
18 | WANG H F, WANG Z F, DU M N, et al. Score-CAM: score-weighted visual explanations for convolutional neural networks[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. Piscataway: IEEE, 2020:111-119. 10.1109/cvprw50498.2020.00020 |
19 | DESAI S, RAMASWAMY H G. Ablation-CAM: visual explanations for deep convolutional network via gradient-free localization[C]// Proceedings of the 2020 IEEE Winter Conference on Applications of Computer Vision. Piscataway: IEEE, 2020: 972-980. 10.1109/wacv45572.2020.9093360 |
20 | DIMOPOULOS Y, BOURRET P, LEK S. Use of some sensitivity criteria for choosing networks with good generalization ability[J]. Neural Processing Letters, 1995, 2(6): 1-4. 10.1007/bf02309007 |
21 | RUCK D W, ROGERS S K, KABRISKY M. Feature selection using a multilayer perceptron[J]. Journal of Neural Network Computing, 1990, 2(2): 40-48. |
22 | KOIKE K, MATSUDA S. New indices for characterizing spatial models of ore deposits by the use of a sensitivity vector and an influence factor[J]. Mathematical Geology, 2006, 38(5): 541-564. 10.1007/s11004-006-9030-3 |
23 | DOMBI G W, NANDI P, SAXE J M, et al. Prediction of rib fracture injury outcome by an artificial neural network[J]. The Journal of Trauma: Injury, Infection, and Critical Care, 1995, 39(5): 915-921. 10.1097/00005373-199511000-00016 |
24 | KOH P W, LIANG P. Understanding black-box predictions via influence functions[C]// Proceedings of the 34th International Conference on Machine Learning. New York: JMLR.org, 2017: 1885-1894. |
25 | RIBEIRO M T, SINGH S, GUESTRIN C. “Why should I trust you?”: explaining the predictions of any classifier[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2016: 1135-1144. 10.1145/2939672.2939778 |
26 | HUI L Y W, SOH D W. Deep interpretation with sign separated and contribution recognized decomposition[C]// Proceedings of the 2021 International Work-Conference on Artificial Neural Networks, LNCS 12861. Cham: Springer, 2021: 395-406. |
27 | GOWAL S, DVIJOTHAM K, STANFORTH R, et al. Scalable verified training for provably robust image classification[C]// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2019: 4841-4850. 10.1109/iccv.2019.00494 |
28 | PEZESHKPOUR P, TIAN Y F, SINGH S. Investigating robustness and interpretability of link prediction via adversarial modifications[C]// Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). Stroudsburg, PA: Association for Computational Linguistics, 2019: 3336-3347. 10.18653/v1/n19-1337 |
29 | MOSHKOVITZ M, YANG Y Y, CHAUDHURI K. Connecting interpretability and robustness in decision trees through separation[C]// Proceedings of the 38th International Conference on Machine Learning. New York: JMLR.org, 2021: 7839-7849. |
30 | KOO P K, PLOENZKE M. Improving representations of genomic sequence motifs in convolutional networks with exponential activations[J]. Nature Machine Intelligence, 2021, 3(3): 258-266. 10.1038/s42256-020-00291-x |
31 | CHEN Z K, JIN S, LIU R Z, et al. A deep non-negative matrix factorization model for big data representation learning[J]. Frontiers in Neurorobotics, 2021, 15: No.701194. 10.3389/fnbot.2021.701194 |
32 | XU Z Q J, ZHANG Y Y, XIAO Y Y. Training behavior of deep neural network in frequency domain[C]// Proceedings of the 2019 International Conference on Neural Information Processing, LNCS 11953. Cham: Springer, 2019: 264-274. |
33 | XU Z Q J, ZHANG Y Y, LUO T, et al. Frequency principle: Fourier analysis sheds light on deep neural networks[J]. Communications in Computational Physics, 2020, 28(5): 1746-1767. 10.4208/cicp.oa-2020-0085 |
34 | XU Z Q J, ZHOU H X. Deep frequency principle towards understanding why deeper learning is faster[C]// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2021:10541-10550. 10.1609/aaai.v35i12.17261 |
35 | ZHANG Y Y, LUO T, MA Z, et al. A linear frequency principle model to understand the absence of overfitting in neural networks[J]. Chinese Physics Letters, 2021, 38(3): No.038701 . 10.1088/0256-307x/38/3/038701 |
36 | ZHANG Y Y, XU Z Q J, LUO T, et al. Explicitizing an implicit bias of the frequency principle in two-layer neural networks[EB/OL]. (2019-05-24) [2021-08-10].. 10.1088/0256-307x/38/3/038701 |
37 | LUO T, MA Z, XU Z Q J, et al. Theory of the frequency principle for general deep neural networks[J]. CSIAM Transactions on Applied Mathematics, 2021, 2(3): 484-507. 10.4208/csiam-am.so-2020-0005 |
38 | WANG H H, WU X D, HUANG Z Y, et al. High-frequency component helps explain the generalization of convolutional neural networks[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 8681-8691. 10.1109/cvpr42600.2020.00871 |
39 | TISHBY N, PEREIRA F C, BIALEK W. The information bottleneck method[EB/OL]. (2000-04-24) [2021-08-10].. |
40 | TISHBY N, ZASLAVSKY N. Deep learning and the information bottleneck principle[C]// Proceedings of the 2015 IEEE Information Theory Workshop. Piscataway: IEEE, 2015: 1-5. 10.1109/itw.2015.7133169 |
41 | SHWARTZ-ZIV R, TISHBY N. Opening the black box of deep neural networks via information[EB/OL]. (2017-04-29) [2021-08-10].. |
42 | ACHILLE A, SOATTO S. Emergence of invariance and disentanglement in deep representations[J]. Journal of Machine Learning Research, 2018, 19: 1-34. 10.1109/ita.2018.8503149 |
43 | PIMENTEL T, VALVODA J, MAUDSLAY R H, et al. Information theoretic probing for linguistic structure[C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: Association for Computational Linguistics, 2020: 4609-4622. 10.18653/v1/2020.acl-main.420 |
44 | HEWITT J, LIANG P. Designing and interpreting probes with control tasks[C]// Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg, PA: Association for Computational Linguistics, 2019: 2733-2743. 10.18653/v1/d19-1275 |
45 | BANG S, XIE P T, LEE H, et al. Explaining a black-box by using a deep variational bottleneck approach[C]// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2021: 11396-11404. 10.1609/aaai.v35i13.17358 |
46 | SABOUR S, FROSST N, HINTON G E. Dynamic routing between capsules[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2017: 3859-3869. |
47 | WU T F, SUN W, LI X L, et al. Towards interpretable R-CNN by unfolding latent structures[EB/OL]. (2018-09-06) [2021-08-10].. |
48 | CHEN X, DUAN Y, HOUTHOOFT R, et al. InfoGAN: interpretable representation learning by information maximizing generative adversarial nets[C]// Proceedings of the 30th International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2016: 2180-2188. |
49 | GU J D. Interpretable graph capsule networks for object recognition[C]// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2021: 1469-1477. 10.1609/aaai.v35i2.16237 |
50 | DAUBECHIES I, DEFRISE M, DE MOL C. An iterative thresholding algorithm for linear inverse problems with a sparsity constraint[J]. Communications on Pure and Applied Mathematics, 2004, 57(11): 1413-1457. 10.1002/cpa.20042 |
51 | ZHOU J T Y, DI K, DU J W, et al. SC2Net: sparse LSTMs for sparse coding[C]// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2018: 4588-4595. 10.1609/aaai.v32i1.11721 |
52 | GREGOR K, LeCUN Y. Learning fast approximations of sparse coding[C]// Proceedings of the 27th International Conference on Machine Learning. Madison, WI: Omnipress, 2010: 399-406. |
53 | LIU J L, CHEN X H, WANG Z Y, et al. ALISTA: analytic weights are as good as learned weights in LISTA[EB/OL]. [2021-08-10].. |
54 | ZHENG S, JAYASRMANA S, ROMERA-PAREDES B, et al. Conditional random fields as recurrent neural networks[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 1529-1537. 10.1109/iccv.2015.179 |
55 | WANG Z Y, LING Q, HUANG T S. Learning deep encoders[C]// Proceedings of the 30th AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2016: 2194-2200. 10.1609/aaai.v30i1.10198 |
56 | ZUO W M, REN D W, ZHANG D, et al. Learning iteration-wise generalized shrinkage-thresholding operators for blind deconvolution[J]. IEEE Transactions on Image Processing, 2016, 25(4): 1751-1764. |
57 | E W. A proposal on machine learning via dynamical systems[J]. Communications in Mathematics and Statistics, 2017, 5(1): 1-11. 10.1007/s40304-017-0103-z |
58 | LU Y P, ZHONG A X, LI Q Z, et al. Beyond finite layer neural networks: bridging deep architectures and numerical differential equations[C]// Proceedings of the 35th International Conference on Machine Learning. New York: JMLR.org, 2018: 3276-3285. |
59 | CHEN R T Q, RUBANOVA Y, BETTENCOURT J, et al. Neural ordinary differential equations[C]// Proceedings of the 32nd International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2018: 6572-6583. |
60 | GRATHWOHL W, CHEN R T Q, BETTENCOURT J, et al. FFJORD: free-form continuous dynamics for scalable reversible generative models[EB/OL]. (2018-10-22) [2021-08-10].. |
61 | LIN Z, ZHANG W, TANG X. Designing partial differential equations for image processing by combining differential invariants[R]. Beijing: Microsoft Research Asia, 2009: 1-26. |
62 | HABER E, RUTHOTTO L. Stable architectures for deep neural networks[J]. Inverse Problems, 2017, 34(1): No.014004. 10.1088/1361-6420/aa9a90 |
63 | RUTHOTTO L, HABER E. Deep neural networks motivated by partial differential equations[J]. Journal of Mathematical Imaging and Vision, 2020, 62(3): 352-364. 10.1007/s10851-019-00903-1 |
64 | ZHANG C Z, LIU A S, LIU X L, et al. Interpreting and improving adversarial robustness of deep neural networks with neuron sensitivity[J]. IEEE Transactions on Image Processing, 2021, 30: 1291-1304. 10.1109/tip.2020.3042083 |
65 | CHEN H G, ZHANG H, BONING D, et al. Robust decision trees against adversarial examples[C]// Proceedings of the 36th International Conference on Machine Learning. New York: JMLR.org, 2019: 1122-1131. |
66 | USTUN B, RUDIN C. Learning optimized risk scores[J]. Journal of Machine Learning Research, 2019, 20: 1-75. |
67 | XIANG L Y, ZHANG H, MA H T, et al. Interpretable complex-valued neural networks for privacy protection[EB/OL]. (2020-01-14) [2021-08-10].. |
68 | 郭文博,徐军. 利用LEMNA解释深度学习在网络安全的应用[J]. 中国教育网络, 2019, 2(3): 40-43. |
GUO W B, XU J. Explain the application of deep learning to network security with LEMNA[J]. China Education Network, 2019, 2(3): 40-43. | |
69 | XU X, ZHENG Q H, YAN Z, et al. Interpretation-enabled software reuse detection based on a multi-level birthmark model[C]// Proceedings of the IEEE/ACM 43rd International Conference on Software Engineering. Piscataway: IEEE, 2021: 873-884. 10.1109/icse43902.2021.00084 |
70 | LU Y C, DONG R H, SMYTH B. Why I like it: multi-task learning for recommendation and explanation[C]// Proceedings of the 12th ACM Conference on Recommender Systems. New York: ACM, 2018: 4-12. 10.1145/3240323.3240365 |
71 | WANG X, HE X N, FENG F L, et al. TEM: tree-enhanced embedding model for explainable recommendation[C]// Proceedings of the 2018 World Wide Web Conference. Republic and Canton of Geneva: International World Wide Web Conferences Steering Committee, 2018: 1543-1552. 10.1145/3178876.3186066 |
72 | MA J X, ZHOU C, CUI P, et al. Learning disentangled representations for recommendation[C/OL]// Proceedings of the 33rd Conference on Neural Information Processing Systems. [2021-08-10].. |
73 | YE L, YANG Y, ZENG J X. An interpretable mechanism for personalized recommendation based on cross feature[J]. Journal of Intelligent and Fuzzy Systems, 2021, 40(5): 9787-9798. 10.3233/jifs-202308 |
74 | BIFFI C, OKTAY O, TARRONI G, et al. Learning interpretable anatomical features through deep generative models: application to cardiac remodeling[C]// Proceedings of the 2018 International Conference on Medical Image Computing and Computer-Assisted Intervention, LNCS 11071. Cham: Springer, 2018: 464-471. |
75 | SCHAUMBERG A J, JUAREZ-NICANOR W C, CHOUDHURY S J, et al. Interpretable multimodal deep learning for real-time pan-tissue pan-disease pathology search on social media[J]. Modern Pathology, 2020, 33(11): 2169-2185. 10.1038/s41379-020-0540-1 |
76 | WANG L F, NIE R, YU Z Y, et al. An interpretable deep-learning architecture of capsule networks for identifying cell-type gene expression programs from single-cell RNA-sequencing data[J]. Nature Machine Intelligence, 2020, 2(11): 693-703. 10.1038/s42256-020-00244-4 |
77 | PINTELAS E, LIASKOS M, LIVIERIS I E, et al. A novel explainable image classification framework: case study on skin cancer and plant disease prediction[J]. Neural Computing and Applications, 2021, 33(22): 15171-15189. 10.1007/s00521-021-06141-0 |
78 | WU L W, RAO Y, ZHAO Y Q, et al. DTCA: decision tree-based co-attention networks for explainable claim verification[C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: Association for Computational Linguistics, 2020: 1024-1035. 10.18653/v1/2020.acl-main.97 |
79 | LU Y J, LI C T. GCAN: graph-aware co-attention networks for explainable fake news detection on social media[C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA: Association for Computational Linguistics, 2020: 505-514. 10.18653/v1/2020.acl-main.48 |
80 | NI L, CHEN Y W, DE BRUJIN O. Towards understanding socially influenced vaccination decision making: an integrated model of multiple criteria belief modelling and social network analysis[J]. European Journal of Operational Research, 2021, 293(1): 276-289. 10.1016/j.ejor.2020.12.011 |
[1] | 秦璟, 秦志光, 李发礼, 彭悦恒. 基于概率稀疏自注意力神经网络的重性抑郁疾患诊断[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2970-2974. |
[2] | 王熙源, 张战成, 徐少康, 张宝成, 罗晓清, 胡伏原. 面向手术导航3D/2D配准的无监督跨域迁移网络[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2911-2918. |
[3] | 黄云川, 江永全, 黄骏涛, 杨燕. 基于元图同构网络的分子毒性预测[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2964-2969. |
[4] | 潘烨新, 杨哲. 基于多级特征双向融合的小目标检测优化模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2871-2877. |
[5] | 李顺勇, 李师毅, 胥瑞, 赵兴旺. 基于自注意力融合的不完整多视图聚类算法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2696-2703. |
[6] | 刘禹含, 吉根林, 张红苹. 基于骨架图与混合注意力的视频行人异常检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2551-2557. |
[7] | 顾焰杰, 张英俊, 刘晓倩, 周围, 孙威. 基于时空多图融合的交通流量预测[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2618-2625. |
[8] | 石乾宏, 杨燕, 江永全, 欧阳小草, 范武波, 陈强, 姜涛, 李媛. 面向空气质量预测的多粒度突变拟合网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2643-2650. |
[9] | 杨莹, 郝晓燕, 于丹, 马垚, 陈永乐. 面向图神经网络模型提取攻击的图数据生成方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2483-2492. |
[10] | 吴筝, 程志友, 汪真天, 汪传建, 王胜, 许辉. 基于深度学习的患者麻醉复苏过程中的头部运动幅度分类方法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2258-2263. |
[11] | 李欢欢, 黄添强, 丁雪梅, 罗海峰, 黄丽清. 基于多尺度时空图卷积网络的交通出行需求预测[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2065-2072. |
[12] | 张郅, 李欣, 叶乃夫, 胡凯茜. 基于暗知识保护的模型窃取防御技术DKP[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2080-2086. |
[13] | 赵亦群, 张志禹, 董雪. 基于密集残差物理信息神经网络的各向异性旅行时计算方法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2310-2318. |
[14] | 徐松, 张文博, 王一帆. 基于时空信息的轻量视频显著性目标检测网络[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2192-2199. |
[15] | 孙逊, 冯睿锋, 陈彦如. 基于深度与实例分割融合的单目3D目标检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2208-2215. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||