volume | PIER Journals

Abstract

Small sample sizes in terahertz (THz) spectroscopy constrain the generalization of machine learning classifiers, motivating data augmentation as a low-cost remedy. However, researchers rarely evaluated the effectiveness of augmentation strategies across diverse classifier architectures. This study cross-evaluates five augmentation strategies (Gaussian noise injection, amplitude scaling, baseline drift, a combined version of the three perturbations, and intra-class Mixup interpolation) on a five-material THz insulating material dataset (129 spectra). Eleven classifiers spanning gradient boosting, bagging ensembles, sequential boosting, kernel methods, linear models, generative models, and neural networks were benchmarked under each augmentation strategy plus a no-augmentation baseline control, yielding 66 independent experiments (6 conditions × 11 models), with hyperparameters tuned via 5-fold stratified cross-validation. On the single fixed partition of the main experiment, the interaction appears large: Mixup improves multi-layer perceptron macro F1 by 83.7% (from 0.416 to 0.765) while AdaBoost degrades by 56.9% (from 0.743 to 0.321), and the best combination, Extra Trees with baseline drift, reaches macro F1 of 0.871. Because such effect sizes may depend on the partition drawn, every cell was reevaluated across 10 independent stratified splits with hyperparameters held fixed. Single-partition magnitudes prove substantially inflated, and the two extreme cases diverge: the MLP gain persists in direction but shrinks to +18.0% (paired ΔF1 = +0.099 ± 0.133), whereas the AdaBoost degradation does not reproduce (+0.064 ± 0.119, 5 of 10 splits negative), its apparent size traceable to a baseline of 0.743 that no repeated split recovered (mean 0.540). What survives is the family structure rather than the individual percentages: augmentation helps neural network (+0.110), kernel (+0.042), and bagging (+0.021 to +0.026) classifiers and harms linear-generative (-0.087) and generative (-0.034) ones. Baseline drift remains the most consistent strategy, its mean gain falling from 0.038 to 0.011. These findings demonstrate that augmentation strategy selection must be guided by classifier architecture, and that effect sizes from a single small-sample partition should be treated as upper bounds. The defensible framework is at the family level: augment neural network, kernel, and bagging classifiers, and avoid augmenting linear-generative and generative ones. Independent-dataset validation is required before generalizing these recommendations.

1. Nsengiyumva, Walter, et al., "Sensing and nondestructive testing applications of terahertz spectroscopy and imaging systems: State-of-the-art and state-of-the-practice," IEEE Transactions on Instrumentation and Measurement, Vol. 72, 1-83, 2023.
doi:10.1109/tim.2023.3318676 Google Scholar

2. Yu, Pengpeng, Chaoping Shen, Wenhui Zhu, Wenya Zhang, Junhui Cheng, and Jinxiu Song, "Recent developments and applications of terahertz spectroscopy in food analysis," Biosensors, Vol. 15, No. 10, 677, October 2025.
doi:10.3390/bios15100677 Google Scholar

3. Vera, E. R., et al., "Advances in THz spectroscopy: A systematic review of material characterization applications," ASEAN Journal of Science and Engineering, Vol. 6, No. 2, 173-208, 2026.
doi:10.17509/ajse.v6i2.90075 Google Scholar

4. Cheng, L., L. Wang, H. Mei, Z. Guan, and F. Zhang, "Research of nondestructive methods to test defects hidden within composite insulators based on THz time-domain spectroscopy technology," IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 23, No. 4, 2126-2133, August 2016.
doi:10.1109/tdei.2016.7556487 Google Scholar

5. Nsengiyumva, Walter, et al., "Terahertz spectroscopic study of optical and dielectric properties of typical electrical insulation materials," Optical Materials, Vol. 123, 111837, January 2022.
doi:10.1016/j.optmat.2021.111837 Google Scholar

6. Helal, Sara, Hadi Sarieddeen, Hayssam Dahrouj, Tareq Y. Al-Naffouri, and Mohamed-Slim Alouini, "Signal processing and machine learning techniques for terahertz sensing: An overview," IEEE Signal Processing Magazine, Vol. 39, No. 5, 42-62, September 2022.
doi:10.1109/msp.2022.3183808 Google Scholar

7. Cacciari, Ilaria and Anedio Ranfagni, "Unlocking Terahertz technology with machine learning: A comprehensive review," Journal of Applied Physics, Vol. 137, No. 22, June 2025.
doi:10.1063/5.0272428 Google Scholar

8. Kristensen, Mathias Hedegaard, Paweł Piotr Cielecki, and Esben Skovsen, "Classification of non-referenced continuous-wave terahertz reflection spectra for remote material identification," Infrared Physics & Technology, Vol. 140, 105420, August 2024.
doi:10.1016/j.infrared.2024.105420 Google Scholar

9. Cortes, Corinna and Vladimir Vapnik, "Support-vector networks," Machine Learning, Vol. 20, No. 3, 273-297, September 1995.
doi:10.1007/bf00994018 Google Scholar

10. Beleites, Claudia, Ute Neugebauer, Thomas Bocklitz, Christoph Krafft, and Jürgen Popp, "Sample size planning for classification models," Analytica Chimica Acta, Vol. 760, 25-33, January 2013.
doi:10.1016/j.aca.2012.11.007 Google Scholar

11. Shorten, Connor and Taghi M. Khoshgoftaar, "A survey on image data augmentation for deep learning," Journal of Big Data, Vol. 6, No. 1, 60, July 2019.
doi:10.1186/s40537-019-0197-0 Google Scholar

12. Yang, Yu, et al., "NIR-GAN: A spectral data augmentation framework for medicine-food homologous herb identification," Journal of Food Composition and Analysis, Vol. 148, 108328, December 2025.
doi:10.2139/ssrn.5344181 Google Scholar

13. Chen, Tiejun and Sung-June Baek, "Library-based Raman spectral identification using multi-input hybrid ResNet," ACS Omega, Vol. 8, No. 40, 37482-37489, September 2023.
doi:10.1021/acsomega.3c05780 Google Scholar

14. Zhang, Hongyi, Moustapha Cisse, Yann N. Dauphin, and David Lopez-Paz, "Mixup: Beyond empirical risk minimization," arXiv:1710.09412, 392-393, 2017.
doi:10.48550/arXiv.1710.09412 Google Scholar

15. Fukunaga, Kaori and Marcello Picollo, "Terahertz spectroscopy applied to the analysis of artists’ materials," Applied Physics A, Vol. 100, No. 3, 591-597, March 2010.
doi:10.1007/s00339-010-5643-y Google Scholar

16. Breiman, Leo, "Random forests," Machine Learning, Vol. 45, No. 1, 5-32, October 2001.
doi:10.1023/a:1010933404324 Google Scholar

17. Geurts, Pierre, Damien Ernst, and Louis Wehenkel, "Extremely randomized trees," Machine Learning, Vol. 63, No. 1, 3-42, March 2006.
doi:10.1007/s10994-006-6226-1 Google Scholar

18. Ke, G., et al., "LightGBM: A highly efficient gradient boosting decision tree," Advances in Neural Information Processing Systems, Vol. 30, 2017.

19. Chen, Tianqi and Carlos Guestrin, "XGBoost: A scalable tree boosting system," Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794, August 2016.
doi:10.1145/2939672.2939785

20. Prokhorenkova, L., G. Gusev, A. Vorobev, A. V. Dorogush, and A. Gulin, "CatBoost: Unbiased boosting with categorical features," Advances in Neural Information Processing Systems, Vol. 31, 2018.

21. Freund, Yoav and Robert E. Schapire, "A decision-theoretic generalization of on-line learning and an application to boosting," Journal of Computer and System Sciences, Vol. 55, No. 1, 119-139, August 1997.
doi:10.1006/jcss.1997.1504 Google Scholar

22. Cox, D. R., "The regression analysis of binary sequences," Journal of the Royal Statistical Society Series B: Statistical Methodology, Vol. 20, No. 2, 215-232, July 1958.
doi:10.1111/j.2517-6161.1958.tb00292.x Google Scholar

23. Fisher, R. A., "The use of multiple measurements in taxonomic problems," Annals of Eugenics, Vol. 7, No. 2, 179-188, September 1936.
doi:10.1111/j.1469-1809.1936.tb02137.x Google Scholar

24. Rumelhart, David E., Geoffrey E. Hinton, and Ronald J. Williams, "Learning representations by back-propagating errors," Nature, Vol. 323, No. 6088, 533-536, October 1986.
doi:10.1038/323533a0 Google Scholar

25. Zhang, H., et al., "The optimality of naive Bayes," Aa, Vol. 1, No. 2, 3, 2004. Google Scholar

26. LeCun, Y., L. Bottou, Y. Bengio, and P. Haffner, "Gradient-based learning applied to document recognition," Proceedings of the IEEE, Vol. 86, No. 11, 2278-2324, 1998.
doi:10.1109/5.726791 Google Scholar

27. Vaswani, A., et al., "Attention is all you need," Advances in Neural Information Processing Systems, Vol. 30, 2017.

28. Goodfellow, I., et al., "Generative adversarial nets," Advances in Neural Information Processing Systems, Vol. 27, 2014.

Dr. Zhangchong Ye

Dr. Lifeng Zhu

Dr. Luoyi Li

Dr. Jiangyang Zhan

Dr. Lei Zhang

Dr. Jie Cai

Dr. Jing Shen

Dr. Chong Tian

Dr. Xiaofeng Zhang

Dr. Xianliang Li