A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training
Abstract Deep learning is one of the most popular machine learning methods, and has been used in many applications. However, conventional deep learning is a fully deterministic model that has difficulty reducing data uncertainty while shedding no light on model transparency. To this end, a novel adv...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-05-01
|
| Series: | Complex & Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40747-025-01928-3 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849685829458329600 |
|---|---|
| author | Limin Mao Hangming Shi Yongxin Chou Jinliang Cong Mingli Lu Zekang Bian Suhang Gu |
| author_facet | Limin Mao Hangming Shi Yongxin Chou Jinliang Cong Mingli Lu Zekang Bian Suhang Gu |
| author_sort | Limin Mao |
| collection | DOAJ |
| description | Abstract Deep learning is one of the most popular machine learning methods, and has been used in many applications. However, conventional deep learning is a fully deterministic model that has difficulty reducing data uncertainty while shedding no light on model transparency. To this end, a novel adversarial deep Takagi–Sugeno–Kang (TSK) fuzzy classifier (AD-TSK), which is characterized by its novel stacked structure and inverse-free fast training, is proposed in this paper. After taking the TSK fuzzy classifier as the subclassifier in each layer of AD-TSK, proactive adversarial example learning is designed based on the adversarial attacks on both the features and labels of the inputs for each subclassifier to generate the adversarial TSK fuzzy classifier (ATSK) with increased robustness to noise. Moreover, the generalization capability of AD-TSK can be guaranteed by its novel stacked structure constructed using the dropout-based gradients of each ATSK to update the original inputs along the stacked structure in a layer-by-layer manner. In particular, by sharing the merits of the Schur complement and Sherman–Morrison formula, the consequent parameters of all fuzzy rules in each ATSK can be determined without the matrix inversion operation, and the corresponding optimal number of fuzzy rules can be optimized based on the previous consequent parameters of all fuzzy rules. This can facilitate the inverse-free fast deep learning algorithm of AD-TSK while maintaining interpretability. Extensive experiments on fourteen benchmark datasets from the UCI and KEEL data repositories and the corresponding statistical test results demonstrate the effectiveness of the proposed AD-TSK as well as its inverse-free fast training. The source code of AD-TSK classifier can be downloaded from https://github.com/gusuhang10/AD-TSK . |
| format | Article |
| id | doaj-art-99f2dabfc24b4cb1b07861897701f99f |
| institution | DOAJ |
| issn | 2199-4536 2198-6053 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | Springer |
| record_format | Article |
| series | Complex & Intelligent Systems |
| spelling | doaj-art-99f2dabfc24b4cb1b07861897701f99f2025-08-20T03:22:57ZengSpringerComplex & Intelligent Systems2199-45362198-60532025-05-0111712210.1007/s40747-025-01928-3A novel adversarial deep TSK fuzzy classifier with its inverse-free fast trainingLimin Mao0Hangming Shi1Yongxin Chou2Jinliang Cong3Mingli Lu4Zekang Bian5Suhang Gu6School of Electrical Engineering and Automation, Changshu Institute of TechnologySchool of Intelligent Manufacturing, Changzhou UniversitySchool of Electrical Engineering and Automation, Changshu Institute of TechnologySchool of Electrical Engineering and Automation, Changshu Institute of TechnologySchool of Electrical Engineering and Automation, Changshu Institute of TechnologySchool of AI and Computer Science, Jiangnan UniversitySchool of Electrical Engineering and Automation, Changshu Institute of TechnologyAbstract Deep learning is one of the most popular machine learning methods, and has been used in many applications. However, conventional deep learning is a fully deterministic model that has difficulty reducing data uncertainty while shedding no light on model transparency. To this end, a novel adversarial deep Takagi–Sugeno–Kang (TSK) fuzzy classifier (AD-TSK), which is characterized by its novel stacked structure and inverse-free fast training, is proposed in this paper. After taking the TSK fuzzy classifier as the subclassifier in each layer of AD-TSK, proactive adversarial example learning is designed based on the adversarial attacks on both the features and labels of the inputs for each subclassifier to generate the adversarial TSK fuzzy classifier (ATSK) with increased robustness to noise. Moreover, the generalization capability of AD-TSK can be guaranteed by its novel stacked structure constructed using the dropout-based gradients of each ATSK to update the original inputs along the stacked structure in a layer-by-layer manner. In particular, by sharing the merits of the Schur complement and Sherman–Morrison formula, the consequent parameters of all fuzzy rules in each ATSK can be determined without the matrix inversion operation, and the corresponding optimal number of fuzzy rules can be optimized based on the previous consequent parameters of all fuzzy rules. This can facilitate the inverse-free fast deep learning algorithm of AD-TSK while maintaining interpretability. Extensive experiments on fourteen benchmark datasets from the UCI and KEEL data repositories and the corresponding statistical test results demonstrate the effectiveness of the proposed AD-TSK as well as its inverse-free fast training. The source code of AD-TSK classifier can be downloaded from https://github.com/gusuhang10/AD-TSK .https://doi.org/10.1007/s40747-025-01928-3TSK fuzzy classifierInterpretabilityDeep fuzzy systemDeep learningFast learning |
| spellingShingle | Limin Mao Hangming Shi Yongxin Chou Jinliang Cong Mingli Lu Zekang Bian Suhang Gu A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training Complex & Intelligent Systems TSK fuzzy classifier Interpretability Deep fuzzy system Deep learning Fast learning |
| title | A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training |
| title_full | A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training |
| title_fullStr | A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training |
| title_full_unstemmed | A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training |
| title_short | A novel adversarial deep TSK fuzzy classifier with its inverse-free fast training |
| title_sort | novel adversarial deep tsk fuzzy classifier with its inverse free fast training |
| topic | TSK fuzzy classifier Interpretability Deep fuzzy system Deep learning Fast learning |
| url | https://doi.org/10.1007/s40747-025-01928-3 |
| work_keys_str_mv | AT liminmao anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT hangmingshi anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT yongxinchou anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT jinliangcong anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT minglilu anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT zekangbian anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT suhanggu anoveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT liminmao noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT hangmingshi noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT yongxinchou noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT jinliangcong noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT minglilu noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT zekangbian noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining AT suhanggu noveladversarialdeeptskfuzzyclassifierwithitsinversefreefasttraining |