Interpretable Verification Mechanism for Trustworthy Industrial Large Model in Intelligent Manufacturing

Shuxuan Zhao; Guanqin Zhang; Sichao Liu; Jie Zhang; H.M.N. Dilum Bandara; Ray Y. Zhong; Lihui Wang

doi:10.1016/j.eng.2025.08.023

Engineering ›› DOI: 10.1016/j.eng.2025.08.023

Research

research-article

Interpretable Verification Mechanism for Trustworthy Industrial Large Model in Intelligent Manufacturing

Author information +

History +

PDF (4734KB)

Abstract

The hallucination and black-box nature of Large Models limit their industrial applications. To address these challenges, a verification mechanism built on confidence intervals of Transformer-based output layers is proposed for trustworthy Industrial Large Models (ILMs). Adopting a Vision Transformer (ViT), customized verification operations are incorporated to monitor the forward propagation process, and samples with probability distributions outside confidence intervals exit the network early and are handed over to technicians. Thus, the ViT is more interpretable because only samples within confidence intervals can propagate forward and be output from the ViT. Subsequently, an over-approximation approach is employed to obtain confidence intervals by linearizing the decision boundary of the ViT. The conservative decision boundary serves as the lower bound of confidence intervals, which can provide provable robustness for confidence intervals because the minimum probability of the ground truth is always higher than that of other samples. Finally, a certified training strategy is employed to enhance the robustness of the ViT. Data disturbances with Gaussian noise are generated using a randomized smoothing strategy to augment the data distribution. A smoothed loss function is used to strengthen the robustness of the ViT against data disturbances, thereby enabling greater confidence intervals. The proposed verification mechanism was validated on two public defect datasets. It achieved 99.98% precision for normal samples and approximately 95% precision for defective samples on a fabric defect dataset. It also achieved 99.21% precision and 99.15% F1 score on a wafer defect dataset. Comparative experiments with other Transformer-based models also demonstrated the generalization ability of the proposed verification mechanism.

Keywords

Industrial Large Models / Hallucination / Trustworthy / Industrial defect detection

Cite this article

Download citation ▾

Shuxuan Zhao, Guanqin Zhang, Sichao Liu, Jie Zhang, H.M.N. Dilum Bandara, Ray Y. Zhong, Lihui Wang. Interpretable Verification Mechanism for Trustworthy Industrial Large Model in Intelligent Manufacturing. Engineering DOI:10.1016/j.eng.2025.08.023

登录浏览全文

4963

注册一个新账户忘记密码

References

Publishing order | Descend order by publishing year | Descend order by cited within

[1]	R.Y. Zhong, X. Xu, E. Klotz, S.T. Newman. Intelligent manufacturing in the context of industry 4.0: a review. Engineering, 3 (5) (2017), pp. 616-630.

[2]	K. Wang, R. Tan, Q. Peng, L. Zhang, F. Wang. A systematic problem analysis network for product conceptual design. Comput Ind Eng, 194 (2024), Article 110382.

[3]	R.X. Gao, J. Krüger, M. Merklein, H.C. Möhring, J. Váncza. Artificial Intelligence in manufacturing: state of the art, perspectives, and future directions. CIRP Ann, 73 (2) (2024), pp. 723-749.

[4]	S. Zhao, R.Y. Zhong, C. Xu, J. Wang, J. Zhang. A dynamic inference network (DI-Net) for online fabric defect detection in smart manufacturing. J Intell Manuf, 36 (4) (2024), pp. 2881-2896.

[5]	Y. Gao, L. Gao, X. Li, S. Cao. A hierarchical training-convolutional neural network for imbalanced fault diagnosis in complex equipment. IEEE Trans Industr Inform, 18 (11) (2022), pp. 8138-8145.

[6]	Y. Bai, J. Xie, D. Wang, W. Zhang, C. Li. A manufacturing quality prediction model based on AdaBoost-LSTM with rough knowledge. Comput Ind Eng, 155 (2021), Article 107227.

[7]	A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A.N. Gomez, et al. Attention is all you need, Curran Associates Inc., Long Beach, CA, USA. New York (2017), pp. 6000-6010.

[8]	Achiam J, Adler S, Agarwal S, Ahmad L, Akkaya I, Aleman FL, et al. Gpt-4 technical report. 2023. arXiv:2303.08774.

[9]	Koroteev MV. BERT: a review of applications in natural language processing and understanding. 2021. arXiv:2103.11943.

[10]	Minaee S, Mikolov T, Nikzad N, Chenaghlu M, Socher R, Amatriain X, et al. Large language models: a survey. 2024. arXiv:2402.06196.

[11]	H. Wang, J. Li, H. Wu, E. Hovy, Y. Sun. Pre-trained language models and their applications. Engineering, 25 (2023), pp. 51-65.

[12]	A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, et al. Segment anything, IEEE, Paris, France. Piscataway (2023), pp. 3992-4003.

[13]	P. Gao, S. Geng, R. Zhang, T. Ma, R. Fang, Y. Zhang, et al. Clip-adapter: better vision-language models with feature adapters. Int J Comput Vis, 132 (2) (2024), pp. 581-595.

[14]	Wang C, Liu Y, Guo T, LI D, He T, Li Z, et al. Systems engineering issues for industry applications of large language model. Appl Soft Comput 2024; 151:111165.

[15]	H. Wang, C. Li, Y.F. Li. Large-scale visual language model boosted by contrast domain adaptation for intelligent industrial visual monitoring. IEEE Trans Industr Inform, 20 (12) (2024), pp. 14114-14123.

[16]	Gu Z, Zhu B, Zhu G, Chen Y, Tang M, Wang J. (2024). Anomalygpt: detecting industrial anomalies using large vision-language models. Proc AAAI Conf Artif Intell 2024; 38(3):1932-40.

[17]	S. Zhao, R.Y. Zhong, Y. Jiang, S. Besklubova, J. Tao, L. Yin. Hierarchical spatial attention-based cross-scale detection network for Digital Works Supervision System (DWSS). Comput Ind Eng, 192 (2024), Article 110220.

[18]	T. Wang, P. Zheng, S. Li, L. Wang. Multimodal human-robot interaction for human‐centric smart manufacturing: a survey. Adv Intell Syst, 6 (3) (2024), Article 2300359.

[19]	R. Schaeffer, B. Miranda, S. Koyejo Are emergent abilities of large language models a mirage?, Curran Associates Inc., New Orleans, LA, USA. New York (2023), pp. 55565-55581.

[20]	Rawte V, Sheth A, Das A. A survey of hallucination in large foundation models. 2023. arXiv:2309.05922.

[21]	Bai Z, Wang P, Xiao T, He T, Han Z, Zhang Z, et al. Hallucination of multimodal large language models: a survey. 2024. arXiv:2404.18930.

[22]	Liu H, Xue W, Chen Y, Chen D, Zhao X, Wang K, et al. A survey on hallucination in large vision-language models. 2024. arXiv:2402.00253.

[23]	Y. Li, Y. Du, K. Zhou, J. Wang, W.X. Zhao, J.R. Wen . Evaluating object hallucination in large vision-language models, Association for Computational Linguistics, Singapore. Kerrville (2023), pp. 292-305.

[24]	L. Huang, W. Yu, W. Ma, W. Zhong, Z. Feng, H. Wang, et al. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Trans Inf Syst, 43 (2) (2025), pp. 1-55.

[25]	M. Sivakumar, A.B. Belle, J. Shan, S.K. Khakzad. Prompting GPT-4 to support automatic safety case generation. Expert Syst Appl, 255 (2024), Article 124653.

[26]	Z. Yin.A review of methods for alleviating hallucination issues in large language models Appl Comput Eng, 76 (1) (2024), pp. 258-266.

[27]	Yao JY, Ning KP, Liu ZH, Ning MN, Liu YY, Yuan L. LLM lies: hallucinations are not bugs, but features as adversarial examples. 2023. arXiv:2310.01469.

[28]	H. Wang, C. Li, Y.F. Li, F. Tsung. An intelligent industrial visual monitoring and maintenance framework empowered by large-scale visual and language models. IEEE Trans Ind Cyber-Phys Syst, 2 (2024), pp. 166-175.

[29]	A. Kusiak. Generative artificial intelligence in smart manufacturing. J Intell Manuf, 36 (2024), pp. 1-3.

[30]	H. Wang, M. Liu, W. Shen. Industrial‐generative pre‐trained Transformer for intelligent manufacturing systems. IET Collab Intell Manuf, 5 (2) (2023), Article e12078.

[31]	W. Wang, Z. Chen, X. Chen, J. Wu, X. Zhu, G. Zeng, et al. Visionllm:large language model is also an open-ended decoder for vision-centric tasks, Curran Associates Inc., New Orleans, LA, USA. New York (2023), pp. 61501-61513.

[32]	T. Wang, J. Fan, P. Zheng. An LLM-based vision and language cobot navigation approach for human-centric smart manufacturing. J Manuf Syst, 75 (2024), pp. 299-305.

[33]	C. Gkournelos, C. Konstantinou, S. Makris. An LLM-based approach for enabling seamless human-robot collaboration in assembly. CIRP Ann, 73 (1) (2024), pp. 9-12.

[34]	Alsaqer S, Alajmi S, Ahmad I, Alfailakawi M. J Eng Res The potential of LLMs in hardware design. 2024; In Press.

[35]	H. Liu, Y. Wang, W. Fan, X. Liu, Y. Li, S. Jain, et al. Trustworthy ai: a computational perspective. ACM Trans Intell Syst Technol, 14 (1) (2023), pp. 1-59.

[36]	X. He, W. Huang, C. Lv. Toward trustworthy decision-making for autonomous vehicles: a robust reinforcement learning approach with safety guarantees. Engineering, 33 (2024), pp. 77-89.

[37]	N.K. Kitson, A.C. Constantinou, Z. Guo, Y. Liu, K. Chobtham. A survey of bayesian network structure learning. Artif Intell Rev, 56 (8) (2023), pp. 8721-8814.

[38]	T. Zhou, L. Zhang, T. Han, E.L. Droguett, A. Mosleh, F.T. Chan. An uncertainty-informed framework for trustworthy fault diagnosis in safety-critical applications. Reliab Eng Syst Saf, 229 (2023), Article 108865.

[39]	L. Sun, Y. Huang, H. Wang, S. Wu, Q. Zhang, C. Gao, et al. Trustllm: trustworthiness in large language models., arXiv:2401.05561 (2024).

[40]	X. Huang, W. Ruan, W. Huang, G. Jin, Y. Dong, C. Wu, et al. A survey of safety and trustworthiness of large language models through the lens of verification and validation. Artif Intell Rev, 57 (7) (2024), p. 175.

[41]	M. Fazlyab, M. Morari, G.J. Pappas. Safety verification and robustness analysis of neural networks via quadratic constraints and semidefinite programming. IEEE Trans Automat Contr, 67 (1) (2022), pp. 1-15.

[42]	E. Nasarian, R. Alizadehsani, U.R. Acharya, K.L. Tsui. Designing interpretable ML system to enhance trust in healthcare: a systematic review to proposed responsible clinician-AI-collaboration framework. Inf Fusion, 108 (2024), Article 102412.

[43]	C.V. Goldman, M. Baltaxe, D. Chakraborty, J. Arinez, C.E. Diaz. Interpreting learning models in manufacturing processes: towards explainable AI methods to improve trust in classifier predictions. J Ind Inf Integr, 33 (2023), Article 100439.

[44]	G. Ciravegna, P. Barbiero, F. Giannini, M. Gori, P. Lió, M. Maggini, et al. Logic explained networks. Artif Intell, 314 (2023), Article 103822.

[45]	J. Wang, S. Zhao, C. Xu, J. Zhang, R. Zhong. Brain-inspired interpretable network pruning for smart vision-based defect detection equipment. IEEE Trans Industr Inform, 19 (2) (2023), pp. 1666-1673.

[46]	S. Zhao, R.Y. Zhong, J. Wang, C. Xu, J. Zhang. Unsupervised fabric defects detection based on spatial domain saliency and features clustering. Comput Ind Eng, 185 (2023), Article 109681.

[47]	Y.N. Sun, W. Qin, J.H. Hu, H.W. Xu, P.Z. Sun. A causal model-inspired automatic feature-selection method for developing data-driven soft sensors in complex industrial processes. Engineering, 22 (2023), pp. 82-93.

[48]	A. Puthanveettil Madathil, X. Luo, Q. Liu, C. Walker, R. Madarkar, Y. Cai, et al. Intrinsic and post-hoc XAI approaches for fingerprint identification and response prediction in smart manufacturing processes. J Intell Manuf, 35 (8) (2024), pp. 4159-4180.

[49]	K. Han, Y. Wang, H. Chen, X. Chen, J. Guo, Z. Liu, et al. A survey on vision transformer. IEEE Trans Pattern Anal Mach Intell, 45 (1) (2023), pp. 87-110.

[50]	A. Milan, S. Rezatofighi, R. Garg, A. Dick, I. Reid Data-driven approximations to NP-hard problems, AAAI Press, San Francisco, CA, USA. Washington (2017), pp. 1453-1459.

[51]	J. Wang, C. Xu, Z. Yang, J. Zhang, X. Li. Deformable convolutional networks for efficient mixed-type wafer defect pattern recognition. IEEE Trans Semicond Manuf, 33 (4) (2020), pp. 587-596.

[52]	Liu Z, Lin Y, Cao Y, Hu H, Wei Y, Zhang Z, et al. (2021). Swin transformer:hierarchical vision transformer using shifted windows. In:Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV); 2021 Oct 10-17; Montreal, QC, Canada. Piscataway: IEEE; 2021. p. 9992-10002.

[53]	Tu Z, Talebi H, Zhang H, Yang F, Milanfar P, Bovik A, et al. Maxvit:multi-axis vision transformer. In: AvidanS, BrostowG, CisséM, FarinellaGM, HassnerT, editors. Computer Vision—ECCV 2022: 17 th European Conference. Lecture Notes in Computer Science, vol 13684. Cham: Springer; 2022. p. 459-79.

[54]	G. Zhang, J. Sun, F. Xu, Y. Sui, H.D. Bandara, S. Chen, et al. A tale of two cities: data and configuration variances in robust deep learning. IEEE Internet Comput, 27 (6) (2023), pp. 13-20.

[55]	S. Zhao, S. Liu, Y. Jiang, B. Zhao, Y. Lv, J. Zhang, et al. Industrial foundation models (IFMs) for intelligent manufacturing: a systematic review. J Manuf Syst, 82 (2025), pp. 420-448.