Identification of Complex Mixtures for Raman Spectroscopy Using a Novel Scheme Based on a New Multi-Label Deep Neural Network
Open accessThe DOI matches its Crossref record; 43 of the 45 references it cites were found in a registry or an index.
Abstract
With a noisy environment caused by fluorescence and additive white noise as well as complicated spectra fingerprints, the identification of complex mixture materials remains a significant challenge in Raman spectroscopy application. This paper proposes a new scheme based on a constant wavelet transform (CWT) and a deep network for classifying complex mixture. The scheme first transforms the noisy Raman spectra to a two-dimensional scale map using CWT. Then, the multi-label deep neural network model (MLDNN) is simulated for material classification. The proposed model accelerates the feature extraction and expands the feature graph using the global averaging pooling layer. The Sigmoid function is implemented in the last layer of the model. The MLDNN model was trained, validated, and tested with data collected from the samples prepared from palm oil substances. During the training and validating process, data augmentation is applied to overcome the imbalance of data and enrich Raman spectra's diversity. From the test results, MLDNN successfully surpassed VGG16, VGG19, ResNet50, DenseNet121, InceptionResNetV2, MobileNet50 in average precision, F1 macro averaging and F1 micro averaging, which were 0.9900, 0.9874 and 0.9874, respectively. In Hamming loss, one error, coverage, ranking loss, the MLDNN model reached 0.014, 0.020, 1.738 and 0.000, respectively. The values of other models in Hamming loss, one error, coverage, ranking loss are higher than those of MLDNN. In the comparison of AUC value, MLDNN outperforms other models on each label. The average detection time of the MLDNN is 5.3123 s, which is much faster than those of VGG16, VGG19, ResNet50, DenseNet121, InceptionResNetV2 and MobileNetV2, which are 7.1245, 7.5046, 8.6300, 12.3294, 16.1131 and 6.6451, respectively.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
SMOTE synthetic minority over-sampling technique
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002) · Journal of Artificial Intelligence Research · doi:10.1613/jair.953 · Cited by 32,865 · cited 1×
“The synthetic minority oversampling technique (SMOTE) was proposed to overcome this problem [36].”
MethodInception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi, A. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.11231 · Cited by 4,636 · cited 10×
“A. Result Analysis In this experiment, we use the MLDNN models, which have won in the image recognition competition in recent years, mainly VGG16, VGG19, ResNet50, MobileNetV2, DenseNet121, InceptionResNetV2 were investigated and compared with our model.”
BackgroundInfluentialML-KNN: A lazy learning approach to multi-label learning
Zhang, M.-L., & Zhou, Z.-H. (2007) · Pattern Recognition · doi:10.1016/j.patcog.2006.12.019 · Cited by 3,552 · cited 3×
“Eight indicators, which are F1 micro averaging, F1 macro averaging, hamming loss, ROC, subset accuracy, ranking loss, coverage and one error are listed.”
MethodInfluentialA Review on Multi-Label Learning Algorithms
Zhang, M.-L., & Zhou, Z.-H. (2014) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2013.39 · Cited by 3,321 · cited 2×
“It can be proved from (3) that, T Pj + F Pj + T N j + F N j = p, which is constant [43].”
BackgroundChen, Y., Jiang, H., Li, C., Jia, X., & Ghamisi, P. (2016) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2016.2584107 · Cited by 2,984
Classifier chains for multi-label classification
Read, J., Pfahringer, B., Holmes, G., & Frank, E. (2011) · Machine Learning · doi:10.1007/s10994-011-5256-5 · Cited by 2,289 · cited 1×
“For an unknown sample x ∈ X , the multi-label classifier h(·) predicts h(x) ⊆ γ the sample’s label set [41].”
BackgroundSurface-Enhanced Raman Spectroscopy for Bioanalysis: Reliability and Challenges
Zong, C., Xu, M., Xu, L.-J., Wei, T., Ma, X., Zheng, X.-S., Hu, R., & Ren, B. (2018) · Chemical Reviews · doi:10.1021/acs.chemrev.7b00668 · Cited by 1,934 · cited 1×
“There have been several research works on Raman spectroscopy technology; for example, surfaceenhanced Raman spectroscopy [9], high-temperature Raman spectroscopy [10], resonance Raman spectroscopy [11], confocal micro Raman spectroscopy [12], Fourier transform Raman spectroscopy [13], to name a few.”
BackgroundBaseline correction using adaptive iteratively reweighted penalized least squares
Zhang, Z.-M., Chen, S., & Liang, Y.-Z. (2010) · The Analyst · doi:10.1039/b922045c · Cited by 1,124 · cited 1×
“proposed an adaptive iterative weighted least square method for limit correction [21].”
MethodError bounds for approximations with deep ReLU networks
Yarotsky, D. (2017) · Neural Networks · doi:10.1016/j.neunet.2017.07.002 · Cited by 1,085
DenseNet: Implementing Efficient ConvNet Descriptor Pyramids
Iandola, F., Moskewicz, M. W., Karayev, S., Girshick, R., Darrell, T. J., & Keutzer, K. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1404.1869 · Cited by 629 · cited 4×
“(ResNet50) [24], MobileNetV2 [25], DenseNet121 [26], and InceptionResNetV2 [27] can successfully migrate and being”
InfluentialBaseline correction using asymmetrically reweighted penalized least squares smoothing
Baek, S.-J., Park, A., Ahn, Y.-J., & Choo, J. (2015) · The Analyst · doi:10.1039/c4an01061b · Cited by 477 · cited 2×
“Two baseline correction approaches based on hardware and software designs have been proposed [16]–[18].”
MethodFrom Softmax to Sparsemax: A Sparse Model of Attention and Multi-Label Classification
Martins, A. F. T., & Astudillo, R. F. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1602.02068 · Cited by 431 · cited 2×
“In the experiment, the global average pooling is used instead of the full connection layer, which directly averages the entire feature map, and then input it into the Sigmoid layer to get the probability of tags and mappings [40].”
MethodBiological imaging of chemical bonds by stimulated Raman scattering microscopy
Hu, F., Shi, L., & Min, W. (2019) · Nature Methods · doi:10.1038/s41592-019-0538-0 · Cited by 392
Henry, D. G., Jarvis, I., Gillmore, G., & Stephenson, M. (2019) · Earth-Science Reviews · doi:10.1016/j.earscirev.2019.102936 · Cited by 324
Extremely Large Minibatch SGD: Training ResNet-50 on ImageNet in 15 Minutes
Akiba, T., Suzuki, S., & Fukuda, K. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1711.04325 · Cited by 260 · cited 1×
“Due to the rapid development of artificial intelligence, some models that stand out in the competition such as Visual Geometry Group with 16 layers (VGG16) [24], Visual Geometry Group with 19 layers (VGG19) [25], the 50-layer Residual network (ResNet50) [26], MobileNetV2 [27], DenseNet121 [28], and InceptionResNetV2 […”
BackgroundMa, Y., Du, Y., Chen, Y., Gu, C., Jiang, T., Wei, G., & Zhou, J. (2020) · Chemical Engineering Journal · doi:10.1016/j.cej.2019.122710 · Cited by 243 · cited 1×
“[3], physics [4], polymer [5], biology [6], medicine [7] and geology [8].”
BackgroundBaseline correction for Raman spectra using an improved asymmetric least squares method
He, S., Zhang, W., Liu, L., Huang, Y., He, J., Xie, W., Wu, P., & Du, C. (2014) · Anal. Methods · doi:10.1039/c4ay00068d · Cited by 241
Baseline correction by improved iterative polynomial fitting with automatic threshold
Gan, F., Ruan, G., & Mo, J. (2006) · Chemometrics and Intelligent Laboratory Systems · doi:10.1016/j.chemolab.2005.08.009 · Cited by 221 · cited 1×
“proposed an improved iterative polynomial fitting with automatic threshold for limit correction [20].”
BackgroundDeep learning-based component identification for the Raman spectra of mixtures
Fan, X., Ming, W., Zeng, H., Zhang, Z., & Lu, H. (2019) · The Analyst · doi:10.1039/c8an02212g · Cited by 208 · cited 1×
“proposed a novel approach entitled deep learningbased component identification (DeepCID) and established a convolutional neural network (CNN) model to predict the presence of components in mixtures) [28].”
Fernando, E. Y., McIlroy, S. J., Nierychlo, M., Herbst, F.-A., Petriglieri, F., Schmid, M. C., Wagner, M., Nielsen, J. L., & Nielsen, P. H. (2019) · The ISME Journal · doi:10.1038/s41396-019-0399-7 · Cited by 199 · cited 1×
“[3], physics [4], polymer [5], biology [6], medicine [7] and geology [8].”
BackgroundMalicious Software Classification Using VGG16 Deep Neural Network’s Bottleneck Features
Rezende, E., Ruppert, G., Carvalho, T., Theóphilo, A., Ramos, F. T., & de Geus, P. L. (2018) · Advances in intelligent systems and computing · doi:10.1007/978-3-319-77028-4_9 · Cited by 169 · cited 12×
“A. Result Analysis In this experiment, we use the MLDNN models, which have won in the image recognition competition in recent years, mainly VGG16, VGG19, ResNet50, MobileNetV2, DenseNet121, InceptionResNetV2 were investigated and compared with our model.”
BackgroundInfluentialWavelet Approach for ECG Baseline Wander Correction and Noise Reduction
Zhang, D. (2005) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/iembs.2005.1616642 · Cited by 158
Su, Y., Wu, D., Chen, J., Chen, G., Hu, N., Wang, H., Wang, P., Han, H., Li, G., & Wu, Y. (2019) · Analytical Chemistry · doi:10.1021/acs.analchem.9b02233 · Cited by 147
Noise and background removal in Raman spectra of ancient pigments using wavelet transform
Ramos, P. M., & Ruisánchez, I. (2005) · Journal of Raman Spectroscopy · doi:10.1002/jrs.1370 · Cited by 136
Fd-Mobilenet: Improved Mobilenet with a Fast Downsampling Strategy
Qin, Z., Zhang, Z., Chen, X., Wang, C., & Peng, Y. (2018) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2018.8451355 · Cited by 136
High-Temperature Raman Spectroscopy of Nano-Crystalline Carbon in Silicon Oxycarbide
Rosenburg, F., Ionescu, E., Nicoloso, N., & Riedel, R. (2018) · Materials · doi:10.3390/ma11010093 · Cited by 114 · cited 1×
“There have been several research works on Raman spectroscopy technology; for example, surfaceenhanced Raman spectroscopy [9], high-temperature Raman spectroscopy [10], resonance Raman spectroscopy [11], confocal micro Raman spectroscopy [12], Fourier transform Raman spectroscopy [13], to name a few.”
BackgroundNawrocka, A., Szymańska‐Chargot, M., Miś, A., Wilczewska, A. Z., & Markiewicz, K. H. (2016) · Journal of Agricultural and Food Chemistry · doi:10.1021/acs.jafc.5b05712 · Cited by 103 · cited 1×
“There have been several research works on Raman spectroscopy technology; for example, surfaceenhanced Raman spectroscopy [9], high-temperature Raman spectroscopy [10], resonance Raman spectroscopy [11], confocal micro Raman spectroscopy [12], Fourier transform Raman spectroscopy [13], to name a few.”
BackgroundExposing Computer Generated Images by Eye’s Region Classification via Transfer Learning of VGG19 CNN
Carvalho, T., Rezende, E., Alves, M., Balieiro, F. K. C., & Sovat, R. B. (2017) · International Conference on Machine Learning and Applications · doi:10.1109/icmla.2017.00-47 · Cited by 100 · cited 12×
“A. Result Analysis In this experiment, we use the MLDNN models, which have won in the image recognition competition in recent years, mainly VGG16, VGG19, ResNet50, MobileNetV2, DenseNet121, InceptionResNetV2 were investigated and compared with our model.”
BackgroundInfluentialContinuous cell sorting in a flow based on single cell resonance Raman spectra
McIlvenna, D., Huang, W. E., Davison, P. A., Glidle, A., Cooper, J. M., & Yin, H. (2016) · Lab on a Chip · doi:10.1039/c6lc00251j · Cited by 76 · cited 1×
“There have been several research works on Raman spectroscopy technology; for example, surfaceenhanced Raman spectroscopy [9], high-temperature Raman spectroscopy [10], resonance Raman spectroscopy [11], confocal micro Raman spectroscopy [12], Fourier transform Raman spectroscopy [13], to name a few.”
BackgroundJoint Baseline-Correction and Denoising for Raman Spectra
Liu, H., Zhang, Z., Liu, S., Yan, L., Liu, T., & Zhang, T. (2015) · Applied Spectroscopy · doi:10.1366/14-07760 · Cited by 72 · cited 1×
“commonly use the baseline correction method to remove the fluorescence noise. Though many baseline correction algorithms were proposed, these algorithms can denoise within a certain level [16]– [18], [40]. In this paper, we overcome this by transforming one-dimensional (1-D) noisy Raman sepctrum to a two-dimensional (…”
MethodChe, X., Chen, D., & Mi, J. (2019) · Information Sciences · doi:10.1016/j.ins.2019.10.022 · Cited by 71 · cited 2×
“e expresses how far, on average, we need to move down the rank list of labels so as to cover whole ground true labels of the object. The smaller value of coverage the better performance of algorithm. [52]. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.0595 0.0847 0.1195 0.1038 0.0428 0.3133 0.014 Hamming loss 0 0.05 0.1 0.15 0.…”
BackgroundInfluentialChang, D., Saleh, M. A., Gabriels, J. K., Ismail, H., Goldner, B. G., Willner, J., BELDNER, S. J., Mitra, R. L., John, R. M., & Epstein, L. M. (2020) · Journal of the American College of Cardiology · doi:10.1016/j.jacc.2020.04.032 · Cited by 61 · cited 1×
“[3], physics [4], polymer [5], biology [6], medicine [7] and geology [8].”
BackgroundChiu, L.-D., Ho, S.-H., Shimada, R., Ren, N.-Q., & Ozawa, T. (2017) · Biotechnology for Biofuels · doi:10.1186/s13068-016-0691-y · Cited by 50 · cited 1×
“Within the spectrum resolution limit of the Raman spectrometer, it was found that all of the major characteristic peaks were in good agreement with the values reported in other literatures as enlisted in the TABLE I [30]–[32].”
ResultA Novel Pre-Processing Algorithm Based on the Wavelet Transform for Raman Spectrum
Xi, Y., Li, Y., Duan, Z., & Lu, Y. (2018) · Applied Spectroscopy · doi:10.1177/0003702818789695 · Cited by 50 · cited 1×
“There are several unavoidable noise such as shot noise, dark current noise and readout noise [14], [15] in a Raman detector implemented with charge-coupled devices (CCD) and semiconductor devices.”
BackgroundFast Confocal Raman Imaging Using a 2-D Multifocal Array for Parallel Hyperspectral Detection
Kong, L., Navas-Moreno, M., & Chan, J. W. (2015) · Analytical Chemistry · doi:10.1021/acs.analchem.5b03707 · Cited by 43
Zheng, X., Lv, G., Zhang, Y., Lv, X., Gao, Z., Tang, J., & Mo, J. (2019) · Spectrochimica Acta Part A Molecular and Biomolecular Spectroscopy · doi:10.1016/j.saa.2019.02.063 · Cited by 43 · cited 1×
“3059849 can reveal molecular fingerprints according to vibration information [1].”
BackgroundPhysical Layer Performance of Multi-Band Optical Line Systems Using Raman Amplification
Cantono, M., Ferrari, A., Pilori, D., Virgillito, E., Augé, J. L., & Curri, V. (2018) · Journal of Optical Communications and Networking · doi:10.1364/jocn.11.00a103 · Cited by 36
Mecozzi, M. (2014) · APCBEE Procedia · doi:10.1016/j.apcbee.2014.10.003 · Cited by 29 · cited 1×
“eliminates baseline drift in complex environmental samples based on baseline polynomial fitting estimation [19].”
BackgroundLi, N., Li, X., Zou, Z., Lin, L.-R., & Li, Y. (2011) · The Analyst · doi:10.1039/c0an00751j · Cited by 28 · cited 2×
“Though many baseline correction algorithms were proposed, these algorithms can denoise within a certain level [16]– [18], [40].”
BackgroundMethodMethod for Classifying a Noisy Raman Spectrum Based on a Wavelet Transform and a Deep Neural Network
Pan, L., Pipitsunthonsan, P., Daengngam, C., Channumsin, S., Sreesawet, S., & Chongcheawchamnan, M. (2020) · IEEE Access · doi:10.1109/access.2020.3035884 · Cited by 26 · cited 1×
“In [34], experiments have proved the feasibility of using CWT to transform all the original Raman spectra signals into a 2-D multiresolution scale map for classification tasks.”
MethodSelf-citationRecognition of big data mixed Raman spectra based on deep learning with smartphone as Raman analyzer
Liang, J., & Mu, T. (2019) · Electrophoresis · doi:10.1002/elps.201900302 · Cited by 12 · cited 1×
“In [39], rapid recognition of mixtures in complex environments was realized by establishing a fast Raman analysis model based on deep learning through data training, self‐learning, and parameter optimization.”
MethodAllen, A. C., Mangado, J. R., Adams, S., Flynn, M., Chen, B., & Zhang, J. Z. (2018) · The Journal of Physical Chemistry B · doi:10.1021/acs.jpcb.8b06994 · Cited by 7
Mogilevych, B., Isensee, D., Rangel, J. L. A., Pizzol, C. D., Martinello, V. C. A., Dieamant, G., & Martin, A. A. (2015) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2181016 · Cited by 5
Research works printed but not linked: 1
- kester, 0, Taking the mystery out of the infamous formula ’SNR = 6 02N +1 76 dB ’ and why you should care, 7
Cited by
Deep Learning for Raman Spectroscopy: A Review
Luo, R., Popp, J., & Bocklitz, T. (2022) · Analytica · doi:10.3390/analytica3030020 · Cited by 244
“Besides, Pan and his colleagues even increased the Raman data dimension from 1D to 2D by wavelet transform before classification [49,50].”
Recent Progresses in Machine Learning Assisted Raman Spectroscopy
Qi, Y., Hu, D., Jiang, Y., Wu, Z., Zheng, M., Chen, E. X., Liang, Y., Sadi, M. A., Zhang, K., & Chen, Y. P. (2023) · Advanced Optical Materials · doi:10.1002/adom.202203104 · Cited by 250
Pan, L., Zhang, P., Daengngam, C., Peng, S., & Chongcheawchamnan, M. (2022) · Journal of Raman Spectroscopy · doi:10.1002/jrs.6225 · Cited by 116
Self-citationGoel, R., Chakraborty, S., Awasthi, V., Bhardwaj, V., & Dubey, S. K. (2024) · Sensors and Actuators A Physical · doi:10.1016/j.sna.2024.115555 · Cited by 94
Shi, G.-Y., Wu, H.-P., Luo, S.-H., Lu, X., Ren, B. W., Zhang, Q., Lin, W., Chen, R.-Y., Guo, P., Chen, H., et al. (2023) · Analytical Chemistry · doi:10.1021/acs.analchem.3c01101 · Cited by 73
Metadata
- Authors
- Pan, L., Pipitsunthonsan, P., Daengngam, C., Channumsin, S., Sreesawet, S., & Chongcheawchamnan, M.
- Affiliations
- Prince of Songkla University (TH)
- ORCID
- Venue
- IEEE Sensors Journal
- Published
- 2021-05-01
- Volume
- 21
- Issue
- 9
- Pages
- 10834-10843
- ISSN
- 1530-437X (ISSN-L)
- 1558-1748
- 2379-9153
- OpenAlex
- W3131573026
- Cited by
- 29
- References
- 45
- Citations by year
- 2026: 3 · 2025: 7 · 2024: 6 · 2023: 6 · 2022: 5 · 2021: 2
- FWCI
- 2.16
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Sensors Journal (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Science, Research and Innovation Promotion Fund (1383848)
- National Research Council of Thailand (32302)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Spectroscopy Techniques in Biomedical and Chemical Research
- Physical Sciences › Chemistry › Analytical Chemistry › Spectroscopy and Chemometric Analyses
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
- Keywords
- Raman spectroscopy, palm oil, material classification, additive white noise, data augmentation, class imbalance, feature extraction, palm oil adulteration
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.