Abstract：As a typical structure in deep learning, a fully connected network appears in almost all neural network models. In the quantitative analysis of near-infrared spectroscopy, the number of spectral samples is small, but the dimension of each sample is high. It leads to two problems: if the spectrum is directly input into the network, the number of parameters of the network will be very large, which requires more samples to train the model. Otherwise, the model is prone to over fitting; reducing the dimension of the spectrum before inputting it into the network solves the problem that the number of parameters of the network is too large, but it will lose some information and cannot give full play to the learning ability of the network. According to the characteristics of near-infrared spectrum, a group fully connected near-infrared spectrum quantitative analysis network(GFCN) is proposed. Based on the traditional two-layer fully connected network, the network uses several small fully connected layers to replace the first fully connected layer, which overcomes the disadvantage of too many network parameters caused by a direct input spectrum. The GFCN model was tested with Tecator and IDRC2018 datasets and compared with a fully connected network (FCN) and partial least squares (PLS). The results show that the prediction effect of GFCN is better than that of FCN and PLS on the two datasets. In the case of only a small number of samples participating in the modeling, GFCN can still maintain a high prediction effect. The experimental results show that the GFCN can be used for the quantitative analysis of near-infrared spectrum and adapt to the scene with few samples. It indicates that the proposed model has important research value and good application prospects.
[1] PENG Yan-kun, ZHAO Fang, LI Long, et al(彭彦昆, 赵 芳, 李 龙, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2018, 34(5): 159.
[2] SONG Xue-jian, QIAN Li-li, ZHANG Dong-jie, et al(宋雪健, 钱丽丽, 张东杰, 等). Food Research and Development(食品研究与开发), 2017, 38(12): 197.
[3] Zhu L, Lu S H, Zhang Y H, et al. Vibrational Spectroscopy, 2020, 109: 103071.
[4] Yang Y, Wang X, Zhao X, et al. Analytica Chimica Acta, 2021, 1160: 338453.
[5] Luan X, Huang B, Sedghi S, et al. ISA Transactions, 2018, 81: 46.
[6] Suryakala S V, Prince S. Optical and Quantum Electronics, 2019, 51(8): 1.
[7] Xu S, Zhao Y, Wang M, et al. Geoderma, 2018, 310: 29.
[8] Cook R D, Forzani L. Chemometrics and Intelligent Laboratory Systems, 2021, 213: 104307.
[9] Cui C, Fearn T. Chemometrics and Intelligent Laboratory Systems, 2018, 182: 9.
[10] Ng W, Minasny B, Montazerolghaem M, et al. Geoderma, 2019, 352: 251.
[11] Javadi S H, Munnaf M A, Mouazen A M. Geoderma, 2021, 385: 114851.
[12] LI Mao-gang, YAN Chun-hua, XUE Jia, et al(李茂刚, 闫春华, 薛 佳, 等). Chinese Journal of Analytical Chemistry(分析化学), 2019, 47(12): 1995.
[13] Yang M, Xu D, Chen S, et al. Sensors, 2019, 19(2): 263.
[14] Costa L R, Tonoli G H D, Milagres F R, et al. Carbohydrate Polymers, 2019, 224: 115186.
[15] Shen X, Tian X, Liu T, et al. IEEE Transactions on Neural Networks and Learning Systems, 2017, 29(9): 3926.
[16] Rossi F, Lendasse A, François D, et al. Chemometrics and Intelligent Laboratory Systems, 2006, 80(2): 215.
[17] Krier C, François D, Rossi F, et al. Feature Clustering and Mutual Information for the Selection of Variables in Spectral Data. European Symposium on Artificial Neural Networks, 25-27 April, 2007, Bruges, Belgium. 2007: 157.
