Tricholoma matsutake is an expensive product in the global edible fungus market due to its nutritional and medicinal properties. As the price of T. matsutake increases, some adulterated and fraudulent products have also emerged in the market. It is difficult to detect the fraudulent products with similar shape, and the unfair competition often happens. Discriminant methods combined near infrared spectroscopy with chemometrics analysis have been used in many fields. However, due to the high correlation between the spectral data, it is difficult to construct an effective model using original spectra. In this work, a discriminated model developed by the elastic net algorithm and near infrared spectroscopy is presented to determinate the adulterated and fraudulent products of T. matsutake. First, the difference of protein and aspartic acid contents between T. matsutake and three products with similar shape were analyzed. Then, the information variables selected from near infrared spectroscopy using the elastic net were used to establish quantitative models. And, the prediction performance of developed models was evaluated by using the validation set. Finally, the Monte Carlo experiment based on the test set demonstrated the effectiveness of the proposed method. Compared with least absolute shrinkage and selection operator and partial least square regression models, the developed model has a great prediction accuracy and robustness, which can be served as a new discriminant method for T. matsutake adulteration determination.

中文翻译：

---

基于弹性网信息提取的松茸近红外光谱定量分析

由于其营养和药用特性，松茸在全球食用菌市场上是一种昂贵的产品。随着松茸价格的上涨，市场上也出现了一些掺假冒充的产品。形状相似的造假产品难以发现，不正当竞争现象时有发生。近红外光谱与化学计量学分析相结合的判别方法已被用于许多领域。然而，由于光谱数据之间的高度相关性，很难利用原始光谱构建有效的模型。在这项工作中，提出了一种由弹性网算法和近红外光谱开发的判别模型来确定松茸的掺假和欺诈产品。第一的，分析了松茸与三种形状相似的产品之间蛋白质和天冬氨酸含量的差异。然后，使用弹性网从近红外光谱中选择的信息变量用于建立定量模型。并且，通过使用验证集来评估开发模型的预测性能。最后，基于测试集的蒙特卡罗实验证明了所提出方法的有效性。与最小绝对收缩选择算子和偏最小二乘回归模型相比，所建立的模型具有较高的预测精度和鲁棒性，可作为一种新的松茸掺假判别方法。使用弹性网从近红外光谱中选择的信息变量用于建立定量模型。并且，通过使用验证集来评估开发模型的预测性能。最后，基于测试集的蒙特卡罗实验证明了所提出方法的有效性。与最小绝对收缩选择算子和偏最小二乘回归模型相比，所建立的模型具有较高的预测精度和鲁棒性，可作为一种新的松茸掺假判别方法。使用弹性网从近红外光谱中选择的信息变量用于建立定量模型。并且，通过使用验证集来评估开发模型的预测性能。最后，基于测试集的蒙特卡罗实验证明了所提出方法的有效性。与最小绝对收缩选择算子和偏最小二乘回归模型相比，所建立的模型具有较高的预测精度和鲁棒性，可作为一种新的松茸掺假判别方法。基于测试集的蒙特卡罗实验证明了所提出方法的有效性。与最小绝对收缩选择算子和偏最小二乘回归模型相比，所建立的模型具有较高的预测精度和鲁棒性，可作为一种新的松茸掺假判别方法。基于测试集的蒙特卡罗实验证明了所提出方法的有效性。与最小绝对收缩选择算子和偏最小二乘回归模型相比，所建立的模型具有较高的预测精度和鲁棒性，可作为一种新的松茸掺假判别方法。