An Adaptive Fuzzy-Neural Decision Framework for Uncertainty-Aware Classification under Sensor Drift
Keywords:
Concept shift, Domain adaptation, Electronic nose, Probability calibration, Takagi–Sugeno–Kang inferenceAbstract
Background: Gas-sensor arrays lose accuracy over time because sensor responses drift with ageing, poisoning, ambient conditions, and classifiers trained on early measurements become inaccurate and over-confident.
Objectives: To design and rigorously evaluate an Adaptive Fuzzy-Neural framework with Uncertainty Calibration (AFN-UC) that remains usable after deployment without new labels.
Methodology: AFN-UC combines a label-free, prior-matched standardisation of each incoming batch, Gaussian fuzzification with Takagi–Sugeno–Kang rule activations, a compact neural head, temperature calibration, a cost-based reject rule, and a batch-level drift monitor. It was evaluated on the public 36-month gas-sensor drift archive (13,910 samples) and on 10 simulated instruments, against five static, eight adaptive, and one supervised reference method, with 10 seeds, bootstrap confidence intervals, and Holm-corrected Wilcoxon tests.
Results: On real data, AFN-UC reached a macro-F1 score of 0.643 compared with 0.628 for the best static model, but was not better than strong adaptive baselines; batch-standardised logistic regression scored 0.693. In simulation, it reached 0.979. Prior-matched standardisation was decisive when one gas dominated a batch, raising macro-F1 from 0.297 to 0.485 on real data and from 0.415 to 0.674 in simulation. A cost-based reject rule removed 72 % of real-data errors, and inference cost 13.4 µs per sample.
Conclusion: Label-free, prior-aware adaptation is the key ingredient; the neural head adds no accuracy on real data, and calibration under severe drift remains an open problem.
References
A. Rudnitskaya, “Calibration update and drift correction for electronic noses and tongues,” Frontiers in Chemistry, vol. 6, Sept. 2018.
J. Lu, A. Liu, F. Dong, F. Gu, J. Gama and G. Zhang, “Learning under concept drift: A review,” in IEEE Transactions on Knowledge and Data Engineering, vol. 31, no. 12, pp. 2346–2363, Dec. 2019.
N. Dennler, S. Rastogi, J. Fonollosa, A. van Schaik, and M. Schmuker, “Drift in a popular metal oxide sensor dataset reveals limitations for gas classification benchmarks,” Sensors and Actuators B: Chemical, vol. 361, Jun. 2022.
J. Gawlikowski et al., “A survey of uncertainty in deep neural networks,” Artificial Intelligence Review, vol. 56, pp. 1513–1589, Jul. 2023.
M. Minderer et al., “Revisiting the calibration of modern neural networks,” Advances in Neural Information Processing Systems, 2021.
Y. Ovadia et al., “Can you trust your model’s uncertainty? evaluating predictive uncertainty under dataset shift,” Advances in Neural Information Processing Systems, 2019.
N. Talpur, S. J. Abdulkadir, H. Alhussian, M. H. Hasan, N. Aziz, and A. Bamhdi, “Deep Neuro-Fuzzy System application trends, challenges, and future perspectives: A systematic survey,” Artificial Intelligence Review, vol. 56, pp. 865–913, 2023.
D. Wu, C. -T. Lin, J. Huang and Z. Zeng, “On the functional equivalence of TSK fuzzy systems to neural networks, mixture of experts, CART, and stacking ensemble regression,” in IEEE Transactions on Fuzzy Systems, vol. 28, no. 10, pp. 2570–2580, Oct. 2020.
D. Wu, Y. Yuan, J. Huang and Y. Tan, “Optimize TSK fuzzy systems for regression problems: Minibatch gradient descent with regularization, DropRule, and AdaBound (MBGD-RDA),” in IEEE Transactions on Fuzzy Systems, vol. 28, no. 5, pp. 1003–1015, May 2020.
Y. Zhang et al., “TDACNN: Target-domain-free domain adaptation convolutional neural network for drift compensation in gas sensors,” Sensors and Actuators B: Chemical, vol. 361, Jun. 2022.
W. Zhang et al., “Unsupervised attention-based multisource domain adaptation framework for drift compensation in electronic nose systems,” in IEEE Transactions on Instrumentation and Measurement, vol. 74, pp. 1-16, 2025.
J. Lin and X. Zhan, “Sensor-drift compensation in electronic-nose-based gas recognition using knowledge distillation,” Informatics, vol. 13, no. 1, Jan. 2026.
Y. Li, N. Wang, J. Shi, X. Hou, and J. Liu, “Adaptive batch normalization for practical domain adaptation,” Pattern Recognition, vol. 80, pp. 109–117, Aug. 2018.
D. Wang, E. Shelhamer, S. Liu, B. Olshausen, and T. Darrell, “Tent: Fully test-time adaptation by entropy minimization, arXiv, Jun. 2020.
Y. Zou, Z. Yu, B. V. K. Vijay Kumar, and J. Wang, “Unsupervised domain adaptation for semantic segmentation via class-balanced self-training,” Computer Vision, Oct. 2018, pp. 297–313.
Z. Lipton Y.-X. Wang, and A. Smola, “Detecting and correcting for label shift with black box predictors,” Proceedings of the 35th International Conference on Machine Learning, 2018, pp. 3122–3130.
G. Wilson and D. J. Cook, “A survey of unsupervised deep domain adaptation,” ACM Transactions on Intelligent Systems and Technology (TIST), vol. 11, no. 5, pp. 1–46, Jul. 2020.
A. Alexandari, A. Kundaje, and A. Shrikumar, “Maximum likelihood with bias-corrected calibration is hard-to-beat at label shift adaptation,” Proceedings of the 37th International Conference on Machine Learning, pp. 222–232, 2020.
S. Rabanser, S. Günnemann, and Z. Lipton, “Failing loudly: An empirical study of methods for detecting dataset shift,” Advances in Neural Information Processing Systems, 2019.
D. Hendrycks and T. Dietterich, “Benchmarking neural network robustness to common corruptions and perturbations,” arXiv, Mar. 2019.
M. Kull, M. P. Nieto, M. Kängsepp, T. S. Filho, H. Song, and P. Flach, “Beyond temperature scaling: Obtaining well-calibrated multi-class probabilities with dirichlet calibration,” Advances in Neural Information Processing Systems, 2019.
M. Abdar et al., “A review of uncertainty quantification in deep learning: Techniques, applications and challenges,” Information Fusion, vol. 76, pp. 243–297, Dec. 2021.
Y. Geifman and R. El-Yaniv, “Selectivenet: A deep neural network with an integrated reject option,” Proceedings of the 36th International Conference on Machine Learning, 2019, pp. 2151–2159.
K. Hendrickx, L. Perini, D. van der Plas, W. Meert, and J. Davis, “Machine learning with a reject option: A survey,” Machine Learning, vol. 113, pp. 3073–3110, Mar. 2024.
C. R. Harris et al., “Array programming with NumPy,” Nature, vol. 585, pp. 357–362, Sept. 2020.
P. Virtanen et al. “SciPy 1.0: fundamental algorithms for scientific computing in Python,” Nature Methods, vol. 17, pp. 261–272, Feb. 2020.
