Comparing End-to-End Machine Learning Methods for Spectra Classification
Source abstract
Abstract
In scientific research, spectroscopy and diffraction experimental techniques are widely used and produce huge amounts of spectral data. Learning patterns from spectra is critical during these experiments. This provides immediate feedback on the actual status of the experiment (e.g., time-resolved status of the sample), which helps guide the experiment. The two major spectral changes what we aim to capture are either the change in intensity distribution (e.g., drop or appearance) of peaks at certain locations, or the shift of those on the spectrum. This study aims to develop deep learning (DL) classification frameworks for one-dimensional (1D) spectral time series. In this work, we deal with the spectra classification problem from two different perspectives, one is a general two-dimensional (2D) space segmentation problem, and the other is a common 1D time series classification problem. We focused on the two proposed classification models under these two settings, the namely the end-to-end binned Fully Connected Neural Network (FCNN) with the automatically capturing weighting factors model and the convolutional SCT attention model. Under the setting of 1D time series classification, several other end-to-end structures based on FCNN, Convolutional Neural Network (CNN), ResNets, Long Short-Term Memory (LSTM), and Transformer were explored. Finally, we evaluated and compared the performance of these classification models based on the High Energy Density (HED) spectra dataset from multiple perspectives, and further performed the feature importance analysis to explore their interpretability. The results show that all the applied models can achieve 100% classification confidence, but the models applied under the 1D time series classification setting are superior. Among them, Transformer-based methods consume the least training time (0.449 s). Our proposed convolutional Spatial-Channel-Temporal (SCT) attention model uses 1.269 s, but its self-attention mechanism performed across spatial, channel, and temporal dimensions can suppress indistinguishable features better than others, and selectively focus on obvious features with high separability.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Advanced Chemical Sensor Technologies · Spectroscopy and Chemometric Analyses · Time Series Analysis and Forecasting
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Software Engineering Department, Institute of Informatics, University of Szeged, Dugonics tér 13., 6720 Szeged, Hungary
European XFEL GmbH, Holzkoppel 4., 22869 Schenefeld, Germany
NOMAD HUB Data Center, Humboldt University of Berlin, Zum Großen Windkanal 2, 12489 Berlin, Germany
MTA-SZTE Research Group on Artificial Intelligence, ELKH, Dugonics tér 13., 6720 Szeged, Hungary
Funding
China Scholarship Council
References (67)
Show source-provided references
1. Deep learning for time series classification: A review10.1007/s10618-019-00619-1 ↗
2. Kamath, U., Liu, J., and Whitaker, J. (2019). Deep Learning for NLP and Speech Recognition, Springer.10.1007/978-3-030-14596-5 ↗
3. Deep learning algorithms for human activity recognition using mobile and wearable sensor networks: State of the art and research challenges10.1016/j.eswa.2018.03.056 ↗
4. Deep cybersecurity: A comprehensive overview from neural network and deep learning perspective10.1007/s42979-021-00535-6 ↗
5. Self-attention for raw optical satellite time series classification10.1016/j.isprsjprs.2020.06.006 ↗
6. Liu, J., Yan, J., Wang, L., Huang, L., He, H., and Liu, H. (2021). Remote Sensing Time Series Classification Based on Self-Attention Mechanism and Time Sequence Enhancement. Remote Sens., 13.10.3390/rs13091804 ↗
7. Deep learning spectroscopy: Neural networks for molecular excitation spectra10.1002/advs.201801367 ↗
8. Adversarial nets for baseline correction in spectra processing10.1016/j.chemolab.2021.104317 ↗
9. Deep learning10.1038/nature14539 ↗
10. Edelen, A., Mayes, C., Bowring, D., Ratner, D., Adelmann, A., Ischebeck, R., Snuverink, J., Agapov, I., Kammering, R., and Edelen, J. (2018). Opportunities in machine learning for particle accelerators. arXiv.
11. Nakatsutsumi, M., Tschentscher, T., Cowan, T., Ferrari, A., Schlenvoigt, H.P., Appel, K., Strempfer, J., and Zimmermann, M.V. (2014). Scientific Instrument High Energy Density Physics (HED), European X-ray Free-Electron Laser Facility GmbH.
12. Sun, Y., Brockhauser, S., and Hegedűs, P. (2021). Machine Learning Applied for Spectra Classification. International Conference on Computational Science and Its Applications, Springer.10.1007/978-3-030-87013-3_5 ↗
13. Learning k for knn classification
14. Nondestructive discrimination of pharmaceutical preparations using near-infrared spectroscopy and partial least-squares discriminant analysis10.1080/00032719.2017.1339070 ↗
15. Geographical recognition of Syrah wines by combining feature selection with Extreme Learning Machine10.1016/j.measurement.2018.01.052 ↗
16. Charge state determination of peptide tandem mass spectra using support vector machine (SVM)10.1109/TITB.2010.2040287 ↗
17. Karim, F., Majumdar, H.S., and Darabi, S.C. (2017). Lstm fully convolutional networks for time series classication. arXiv.10.1109/ACCESS.2017.2779939 ↗
18. Lai, G., Chang, W.C., Yang, Y., and Liu, H. (2018, January 8–12). Modeling long-and short-term temporal patterns with deep neural networks. Proceedings of the 41st International ACM SIGIR Conference on Research & Development in Information Retrieval, New York, NY, USA.10.1145/3209978.3210006 ↗
19. Tan, C.W., Bergmeir, C., Petitjean, F., and Webb, G.I. (2020). Monash University, UEA, UCR Time Series Regression Archive. arXiv.
20. Zerveas, G., Jayaraman, S., Patel, D., Bhamidipaty, A., and Eickhoff, C. (2020). A Transformer-based Framework for Multivariate Time Series Representation Learning. arXiv.10.1145/3447548.3467401 ↗
21. Garnot, V.S.F., Landrieu, L., Giordano, S., and Chehata, N. (2019). Satellite image time series classification with pixel-set encoders and temporal self-attention. arXiv.
22. Using deep learning to predict soil properties from regional spectral data10.1016/j.geodrs.2018.e00198 ↗
23. ROCKET: Exceptionally fast and accurate time series classification using random convolutional kernels10.1007/s10618-020-00701-z ↗
24. Hammerla, N.Y., Halloran, S., and Plötz, T. (2016). Deep, convolutional, and recurrent models for human activity recognition using wearables. arXiv.
25. Lipton, Z.C., Kale, D.C., Elkan, C., and Wetzel, R. (2015). Learning to diagnose with LSTM recurrent neural networks. arXiv.
26. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L., and Polosukhin, I. (2017). Attention is all you need. arXiv.
27. Wang, Z., Yan, W., and Oates, T. (2017, January 14–19). Time series classification from scratch with deep neural networks: A strong baseline. Proceedings of the IJCNN 2017: International Joint Conference on Neural Networks, Anchorage, Alaska.10.1109/IJCNN.2017.7966039 ↗
28. Bertasius, G., Wang, H., and Torresani, L. (2021). Is Space-Time Attention All You Need for Video Understanding?. arXiv.
29. Convolutional neural networks: An overview and application in radiology10.1007/s13244-018-0639-9 ↗
30. Spectra data classification with kernel extreme learning machine10.1016/j.chemolab.2019.103815 ↗
31. Chen, Z., Andrejevic, N., Drucker, N., Nguyen, T., Xian, R.P., Smidt, T., Wang, Y., Ernstorfer, R., Tennant, A., and Chan, M. (2021). Machine learning on neutron and x-ray scattering. arXiv.10.1063/5.0049111 ↗
32. Chlorophyll estimation using multi-spectral unmanned aerial system based on machine learning techniques
33. Nearest clusters based partial least squares discriminant analysis for the classification of spectral data10.1016/j.aca.2018.01.023 ↗
34. Phase Mapper: Accelerating Materials Discovery with AI
35. Pelletier, C., Webb, G.I., and Petitjean, F. (2019). Temporal Convolutional Neural Network for the Classification of Satellite Image Time Series. Remote Sens., 11.10.3390/rs11050523 ↗
36. Deep learning based multi-temporal crop classification10.1016/j.rse.2018.11.032 ↗
37. He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27–30). Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA.10.1109/CVPR.2016.90 ↗
38. Multiscale kernel based residual convolutional neural network for motor fault diagnosis under nonstationary conditions10.1109/TII.2019.2941868 ↗
39. Spatial pyramid pooling in deep convolutional networks for visual recognition10.1109/TPAMI.2015.2389824 ↗
40. Yazdanbakhsh, O., and Dick, S. (2019). Multivariate time series classification using dilated convolutional neural network. arXiv.
41. Dilated convolutional neural networks for time series forecasting
42. Rapid identification of pathogenic bacteria using Raman spectroscopy and deep learning10.1038/s41467-019-12898-9 ↗
43. Convolutional neural network based approach for classification of edible oils using low-field nuclear magnetic resonance10.1016/j.jfca.2020.103566 ↗
44. Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks10.1038/s41524-019-0196-x ↗
45. Multivariate LSTM-FCNs for time series classification10.1016/j.neunet.2019.04.014 ↗
46. Gao, S., Ramanathan, A., and Tourassi, G. (2018, January 20). Hierarchical convolutional attention networks for text classification. Proceedings of the Third Workshop on Representation Learning for NLP, Melbourne, Australia.10.18653/v1/W18-3002 ↗
47. Fully convolutional attention network for biomedical image segmentation10.1016/j.artmed.2020.101899 ↗
48. Long short-term memory10.1162/neco.1997.9.8.1735 ↗
49. Cho, K., Van Merriënboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., and Bengio, Y. (2014). Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv.10.3115/v1/D14-1179 ↗
50. Dasgupta, S., and Osogami, T. (2017, January 4–9). Nonlinear dynamic Boltzmann machines for time-series prediction. Proceedings of the AAAI Conference on Artificial Intelligence, San Francisco, CA, USA.10.1609/aaai.v31i1.10806 ↗
51. Temporal pattern attention for multivariate time series forecasting10.1007/s10994-019-05815-0 ↗
52. DuPLO: A DUal view Point deep Learning architecture for time series classificatiOn10.1016/j.isprsjprs.2019.01.011 ↗
53. Detection of pseudo brain tumors via stacked LSTM neural networks using MR spectroscopy signals10.1016/j.bbe.2020.12.003 ↗
54. Discrimination of blood species using Raman spectroscopy combined with a recurrent neural network10.1364/OSAC.416351 ↗
55. He, X., Chen, Y., and Lin, Z. (2021). Spatial-Spectral Transformer for Hyperspectral Image Classification. Remote Sens., 13.10.3390/rs13030498 ↗
56. Wu, N., Green, B., Ben, X., and O’Banion, S. (2020). Deep transformer models for time series forecasting: The influenza prevalence case. arXiv.
57. Ma, J., Shou, Z., Zareian, A., Mansour, H., Vetro, A., and Chang, S.F. (2019). CDSA: Cross-dimensional self-attention for multivariate, geo-tagged time series imputation. arXiv.
58. Towards a universal neural network encoder for time series
59. Convolutional neural networks for time series classification10.21629/JSEE.2017.01.18 ↗
60. Sun, Y., and Brockhauser, S. (2021). Machine Learning Applied for Spectra Classification in XFEL Sciences. Data Sci. J., submitted.10.5334/dsj-2022-015 ↗
61. Nair, V., and Hinton, G.E. (2010, January 21–24). Rectified linear units improve restricted boltzmann machines. Proceedings of the 27th International Conference on International Conference on Machine Learning, Haifa, Israel.
62. Han, J., and Moraga, C. (1995). The influence of the sigmoid function parameters on the speed of backpropagation learning. International Workshop on Artificial Neural Networks, Springer.10.1007/3-540-59497-3_175 ↗
63. Fu, J., Liu, J., Tian, H., Li, Y., Bao, Y., Fang, Z., and Lu, H. (2019, January 15–20). Dual attention network for scene segmentation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, CA, USA.10.1109/CVPR.2019.00326 ↗
64. Dropout: A simple way to prevent neural networks from overfitting
65. A review on the long short-term memory model10.1007/s10462-020-09838-1 ↗
66. Kingma, D.P., and Ba, J. (2014). Adam: A method for stochastic optimization. arXiv.
67. Mersenne twister: A 623-dimensionally equidistributed uniform pseudo-random number generator10.1145/272991.272995 ↗