Method for predicting and identifying whether soy sauce contains yeast extract

Abstract

The invention belongs to the field of analytical chemistry and mathematical modeling, and particularly relates to a method for predicting and identifying whether soy sauce contains yeast extracts, which comprises the following steps: 1) taking different types of soy sauce as modeling samples; 2) carrying out decolorization pretreatment on the modeling sample; (3) carrying out spectrum scanning on the modeling sample by adopting a Raman spectrum technology to obtain original Raman spectrum information; 4) preprocessing the Raman spectrum data obtained in the step 3); 5) sampling for multiple times to establish a sparse logistic regression model, and recording important wave numbers; 6) counting wavenumber importance; 7) selecting an important wave number; 8) using TabPFN to establish a prediction model; (9) predicting and judging whether the yeast extract is added into the soy sauce sample or not by utilizing the established prediction model; compared with the prior art, the method can quickly and accurately identify whether the soy sauce such as light soy sauce and dark soy sauce contains the yeast extract or not, and is simple and quick in identification process, short in analysis time, high in accuracy and high in practicability.

Description

[Technical field]

[0001] The invention belongs to the field of analytical chemistry and mathematical modeling, and specifically is a method for predicting and identifying whether soy sauce contains yeast extract based on Raman spectroscopy technology and Monte Carlo-sparse logistic regression combined with a table probability feedforward network algorithm. [Background Technology]

[0002] Soy sauce is a nutritious condiment with unique flavor and color. It is deeply loved by the Chinese people and is an indispensable part of the common people's table. As people's quality of life gradually improves, consumers' demand for soy sauce shows a trend of high quality development, and they pay more attention to the additive-free nature and taste of soy sauce. Green and organic, delicious taste, and long aftertaste are key indicators. Additive-free soy sauce refers to soy sauce produced without any food additives in the production process, and only raw and auxiliary materials with food ingredient properties are used.

[0003] Yeast extract is a natural condiment made from yeast using modern bio-enzymatic technology. It does not involve artificial chemical synthesis and can be used in additive-free soy sauce. It has been widely used in the field of condiments due to its strong meaty flavor, delicious taste, and green and healthy characteristics. It can produce a flavor enhancement effect that is greater than the sum of its parts when combined with modern brewing technology. Yeast extract can increase the amino acids, peptides, organic acids, and flavor substances in soy sauce, making up for the shortcomings of bland taste and inconsistent flavors caused by short fermentation time, and enhancing the umami flavor and taste of soy sauce.

[0004] In the current market, some manufacturers, in pursuit of short-term profits, falsely add yeast extract and do not add or add artificial synthetic additives to soy sauce to pretend that it has been added with yeast extract. This not only reduces the quality of soy sauce, but also poses a food safety hazard, bringing a negative impact to the soy sauce market.

[0005] Raman spectroscopy is a non-destructive spectroscopic technique that analyzes the chemical composition and structure of a sample by measuring the scattered spectrum after the sample interacts with light. When a laser is irradiated onto a sample, most of the light will scatter at the same frequency (Rayleigh scattering), but a small portion of the light will scatter at a different frequency (Raman scattering). These frequency changes are related to the molecular vibration and rotation in the sample, so they can be used to identify the composition of the sample. However, in actual food testing, due to the complex composition of food, the Raman signals between different compounds will interfere with each other, which also poses a challenge to the use of Raman spectroscopy for complex sample analysis. In recent years, there have been some breakthroughs in the application of Raman spectroscopy combined with machine learning in the field of food testing, and the application of Raman spectroscopy has also attracted more and more attention, which is expected to enable Raman spectroscopy-based analysis and detection technology to be applied to complex food samples.

[0006] The rapid development of science and technology has gradually applied data-driven algorithms represented by deep learning to various fields. Generally, deep learning algorithms have high requirements for data volume and computing hardware. However, the sample capacity that can be used for model training in the field of food analysis is usually small, and there are high requirements for the response speed of the algorithm, which is not suitable for general deep learning algorithms. Tabular Probabilistic Feedforward Network (hereinafter referred to as TabPFN) is a special deep learning method for tabular data. Its main advantages are high efficiency and applicability to small data sets. The architectural design of TabPFN can make full use of the advantages of probabilistic modeling and feedforward neural networks. Raman spectral data is usually tabular data, so TabPFN has great potential in data analysis based on Raman spectroscopy. However, the number of variables (wave number points) of Raman spectroscopy is usually greater than the number of samples, so before using TabPFN, it is necessary to reduce the dimension of Raman spectroscopy data, that is, variable selection.

[0007] Sparse Logistic Regression is a machine learning method that combines logistic regression with sparsity constraints. It is particularly effective when processing high-dimensional data because it can automatically select the most important features, thereby reducing the complexity of the model and improving generalization ability. [Summary of the invention]

[0008] The purpose of the present invention is to solve the above-mentioned shortcomings and provide a method for predicting and identifying whether soy sauce contains yeast extract, which can quickly and accurately identify whether soy sauce such as light soy sauce and dark soy sauce contains yeast extract, and the identification process is simple and fast, the analysis time is short, the accuracy is high, and the practicability is high.

[0009] In order to achieve the above object, a method for predicting and identifying whether soy sauce contains yeast extract is designed, comprising the following steps:

[0010] 1) Using different types of soy sauce as modeling samples;

[0011] 2) performing decolorization pretreatment on the modeling sample;

[0012] 3) using Raman spectroscopy to perform spectral scanning on the modeling sample obtained in step 2) to obtain original Raman spectral information;

[0013] 4) Preprocessing the Raman spectrum data obtained in step 3);

[0014] 5) Multiple sampling to establish a sparse logistic regression model and record important wave numbers;

[0015] 6) Statistical wave number importance;

[0016] 7) Select important wave numbers;

[0017] 8) Use TabPFN to build a prediction model;

[0018] 9) using the prediction model established in step 8) to predict and determine whether yeast extract is added to the soy sauce sample.

[0019] Furthermore, in step 1), the sample is a soy sauce sample containing yeast extract or not containing yeast extract, and the soy sauce sample includes light soy sauce and dark soy sauce.

[0020] Furthermore, in step 2), the step of decolorization pretreatment of the modeling sample is as follows: 2.5 mL of soy sauce sample is transferred into a centrifuge tube, 7.5 mL of water and 4 g of activated carbon are added, vortexed and mixed, ultrasonicated at 40 degrees Celsius for 30 minutes, centrifuged at 9000 rpm for 15 minutes, and the supernatant is filtered using a 0.45 μm polytetrafluoroethylene filter membrane.

[0021] Furthermore, in step 3), the instrument used is an Agilent Resolve handheld Raman analysis spectrometer, and its incident depth is set to 1.5.

[0022] Furthermore, in step 4), the steps of preprocessing the Raman spectral data are as follows: assuming the number of samples to be n, the number of wave number points to be p, and each spectrum to correspond to a label; if yeast extract is added, the spectral label value is 1, otherwise it is 0; setting the total number of Monte Carlo samples N and the sampling probability α.

[0023] Furthermore, in step 5), in a Monte Carlo sampling, samples accounting for α of the total number of samples are selected from all Raman spectral data to establish a sparse logistic regression model; the training data is calculated using the cyclic gradient descent method, and the regression coefficient of the sparse logistic regression model is output, the dimension of the regression coefficient is equal to p, which is consistent with the number of wavenumber points of the Raman spectral data, and the Raman spectral wavenumber represented by the regression coefficient that is not zero is recorded.

[0024] Further, in step 6), step 5) is repeated until the total number of samples reaches N, and the ratio of the number of times each Raman spectrum wave number is not 0 in the regression coefficient of sparse logistic regression after N samplings to N is counted. The larger the ratio, the greater the importance of the wave number.

[0025] Furthermore, in step 7), the wave numbers are sorted in descending order according to their importance, and the number of wave numbers n is selected from large to small, which is consistent with the number of samples.

[0026] Furthermore, in step 8), the sample dimension after processing in steps 1 to 5 is n rows and n columns. At this time, the data is used as the independent variable, and whether yeast extract is added is used as the dependent variable, and TabPFN is used to establish a binary classification prediction model.

[0027] Furthermore, in step 9), for a new soy sauce sample, its Raman spectrum is collected and the spectral variables consistent with step 7) are selected, and then brought into the TabPFN prediction model in step 8) to obtain the prediction result of whether yeast extract is added to the sample.

[0028] Compared with the prior art, the present invention provides a method for predicting and identifying whether yeast extract is contained in soy sauce by using a model established by Monte Carlo-sparse logistic regression combined with a table probability feedforward network algorithm, which is suitable for various types of soy sauces such as light soy sauce, dark soy sauce, brewing, and preparation. The identification process is simple and fast, and the result is accurate, which fills the technical gap in the market and meets the detection needs of the market. In summary, the present invention can quickly and accurately identify whether yeast extract is contained in soy sauces such as light soy sauce and dark soy sauce, and the identification result is consistent with the actual situation, the analysis time is short, the accuracy is high, and the practicability is high, and it is worthy of application and promotion. [Drawings]

[0029] Figure 1 It is a schematic diagram of the process of the present invention;

[0030] Figure 2 It is a schematic diagram of variable selection of the TabPFN model of the present invention. [Specific implementation method]

[0031] The invention belongs to the field of analytical chemistry and mathematical modeling, and specifically is a method for identifying yeast extract in soy sauce based on Raman spectroscopy technology and Monte Carlo-sparse logistic regression combined with a tabular probability feedforward network algorithm. The method collects spectra based on Raman spectroscopy technology, combines sparse logistic regression with Monte Carlo sampling (monte-carlo sampling) for variable selection, and the number of selected variables is less than the number of samples used for subsequent analysis, and then TabPFN is used to perform further modeling analysis on the data.

[0032] The basic principle of the present invention is: decolorize the soy sauce sample, collect Raman spectrum characteristic data, and use Monte Carlo-sparse logistic regression combined with a table probability feedforward network algorithm to predict and identify whether the soy sauce contains yeast extract. The specific steps are as follows:

[0033] 1) Using different types of soy sauce as modeling samples;

[0034] 2) performing decolorization pretreatment on the modeling sample;

[0035] 3) Using Raman spectroscopy to scan the modeling sample and obtain original Raman spectral information;

[0036] 4) Preprocessing the Raman spectrum data obtained in step 3);

[0037] 5) Multiple sampling to establish a sparse logistic regression model and record important wave numbers;

[0038] 6) Statistical wave number importance;

[0039] 7) Select important wave numbers;

[0040] 8) Use TabPFN to build a prediction model;

[0041] 9) Using the prediction model established in step 8), predict and determine whether yeast extract is added to the unknown soy sauce sample.

[0042] In step 1), the sample is a soy sauce sample containing yeast extract or not containing yeast extract. It should be understood that the soy sauce sample includes light soy sauce and dark soy sauce.

[0043] In step 2), the step of decolorizing the modeling sample is as follows: accurately pipette 2.5 mL of soy sauce sample into a centrifuge tube, add 7.5 mL of water and 4 g of activated carbon, vortex mix, ultrasonicate at 40 degrees Celsius for 30 minutes, centrifuge at 9000 rpm for 15 minutes, and filter the supernatant using a 0.45 μm polytetrafluoroethylene filter membrane.

[0044] In step 3), the instrument used is an Agilent Resolve handheld Raman analysis spectrometer, and the incident depth is set to 1.5.

[0045] In step 4), the steps of preprocessing the Raman spectral data are as follows: set the number of samples to n, the number of wave number points to p; each spectrum corresponds to a label; if yeast extract is added, the spectral label value is 1, otherwise it is 0; set the total number of Monte Carlo samples N and the sampling probability α.

[0046] In step 5), in a Monte Carlo sampling, select α samples from all Raman spectral data to establish a sparse logistic regression model; use the cyclic gradient descent method to calculate the training data, output the regression coefficient of the sparse logistic regression model, the dimension of the regression coefficient is equal to p, which is consistent with the number of wavenumber points of the Raman spectral data, and record the Raman spectral wavenumber represented by the regression coefficient that is not zero.

[0047] In step 6), repeat step 5) until the total number of samples reaches N; count the number of times each Raman spectrum wave number is not 0 in the regression coefficient of sparse logistic regression after N samplings to N, and the larger the ratio, the greater the importance of the wave number.

[0048] In step 7), the wave numbers are sorted in descending order by their importance, and the number of wave numbers n is selected from large to small, which is consistent with the number of samples.

[0049] In step 8), the sample dimension after processing from step 1 to step 5 is n rows and n columns. At this time, the data is used as the independent variable, and whether yeast extract is added is used as the dependent variable, and TabPFN is used to establish a binary classification prediction model.

[0050] In step 9), for a new soy sauce sample, its Raman spectrum is collected and the spectral variables consistent with step 7) are selected, and then brought into the TabPFN prediction model in step 8) to obtain the prediction result of whether yeast extract is added to the sample.

[0051] The present invention is further described below in conjunction with specific embodiments:

[0052] Example 1: Identification of yeast extracts in 10 light soy sauces of a certain brand

[0053] (1) 10 soy sauce samples were pre-treated for decolorization. 2.5 mL of the soy sauce sample was accurately transferred into 10 centrifuge tubes, 7.5 mL of water and 4 g of activated carbon were added, vortexed and mixed, and then ultrasonicated at 40 degrees Celsius for 30 minutes; centrifuged at 9000 rpm for 15 minutes, and the supernatant was filtered using a 0.45 μm polytetrafluoroethylene filter membrane.

[0054] (2) The instrument used was an Agilent Resolve handheld Raman spectrometer. Under the condition of an incident depth of 1.5, Raman spectra of 10 light soy sauce sample solutions were collected.

[0055] (3) Select spectral variables consistent with the TabPFN prediction model established in the present invention to process the Raman light of 10 light soy sauce samples, bring them into the TabPFN prediction model, and obtain sample prediction results, "0" means no yeast extract, and "1" means yeast extract. After verification, the prediction results are consistent with the actual results, as shown in the following table.

[0056]

[0057]

[0058] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field and will not be described in detail here. The present invention is not limited to the above-mentioned embodiments, and any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for predicting and identifying whether soy sauce contains yeast extract, characterized in that: The following steps are involved: 1) Using different types of soy sauce as modeling samples; 2) performing decolorization pretreatment on the modeling sample; 3) using Raman spectroscopy to perform spectral scanning on the modeling sample obtained in step 2) to obtain original Raman spectral information; 4) Preprocessing the Raman spectrum data obtained in step 3); 5) Multiple sampling to establish a sparse logistic regression model and record important wave numbers; 6) Statistical wave number importance; 7) Select important wave numbers; 8) Use TabPFN to build a prediction model; 9) using the prediction model established in step 8) to predict and determine whether yeast extract is added to the soy sauce sample.

2. The method according to claim 1, characterized in that: In step 1), the sample is a soy sauce sample containing yeast extract or not containing yeast extract, and the soy sauce sample includes light soy sauce and dark soy sauce.

3. The method according to claim 1, characterized in that In step 2), the step of decolorizing the modeling sample is as follows: 2.5 mL of soy sauce sample is transferred into a centrifuge tube, 7.5 mL of water and 4 g of activated carbon are added, vortexed and mixed, ultrasonicated at 40 degrees Celsius for 30 minutes, centrifuged at 9000 rpm for 15 minutes, and the supernatant is filtered using a 0.45 μm polytetrafluoroethylene filter membrane.

4. The method according to claim 1, characterized in that: In step 3), the instrument used is an Agilent Resolve handheld Raman analysis spectrometer, and the incident depth is set to 1.

5.

5. The method according to claim 1, characterized in that In step 4), the steps of preprocessing the Raman spectral data are as follows: set the number of samples to n, the number of wave number points to p, and each spectrum corresponds to a label; if yeast extract is added, the spectral label value is 1, otherwise it is 0; set the total number of Monte Carlo samples N and the sampling probability α.

6. The method according to claim 5, characterized in that: In step 5), in a Monte Carlo sampling, samples accounting for α of the total number of samples are selected from all Raman spectral data to establish a sparse logistic regression model; The training data is calculated using the cyclic gradient descent method to output the regression coefficient of the sparse logistic regression model. The dimension of the regression coefficient is equal to p, which is consistent with the number of wavenumber points of the Raman spectral data. The Raman spectral wavenumber represented by the coefficient whose regression coefficient is not zero is recorded.

7. The method according to claim 6, characterized in that: In step 6), step 5) is repeated until the total number of samples reaches N, and the ratio of the number of times each Raman spectrum wave number is not 0 in the regression coefficient of sparse logistic regression after N samplings to N is counted. The larger the ratio, the greater the importance of the wave number.

8. The method according to claim 7, characterized in that: In step 7), the wave numbers are sorted in descending order by their importance, and the number of wave numbers n is selected from large to small, which is consistent with the number of samples.

9. The method according to claim 8, characterized in that: In step 8), the sample dimension after processing from step 1 to step 5 is n rows and n columns. At this time, the data is used as the independent variable, and whether yeast extract is added is used as the dependent variable, and TabPFN is used to establish a binary classification prediction model.

10. The method according to claim 9, characterized in that: In step 9), for a new soy sauce sample, its Raman spectrum is collected and the spectral variables consistent with step 7) are selected, and then brought into the TabPFN prediction model in step 8) to obtain the prediction result of whether yeast extract is added to the sample.

Images