Model determination method and device applied to food sensory characteristic identification
By establishing a sensory database and using machine learning models to identify the sensory characteristics of food, the problem of inconsistent evaluation standards caused by individual differences among tasters was solved, and unified evaluation of food sensory characteristics and improved accuracy were achieved.
Patent Information
- Application Number
- CN202511035134.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-25
- Publication Date
- 2025-10-03
Smart Images
Figure CN120741791A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of food detection technology, and more specifically, to a model determination method and device for identifying sensory characteristics of food. Background Art
[0002] With the continuous development of science and technology, food quality analysis has become increasingly important and rigorous.
[0003] In related technologies, the sensory properties of food, a crucial aspect of food quality analysis, typically require multiple expert tasters to manually assess the food's sensory characteristics through tasting. However, due to the subjective influence of different tasters, standardization of evaluation criteria is difficult, leading to individual differences that affect the final evaluation results. Summary of the Invention
[0004] To solve the above technical problems, the embodiments of the present application provide a model determination method, device, computer-readable storage medium and electronic device for identifying sensory characteristics of food.
[0005] According to one aspect of an embodiment of the present application, a model determination method for identifying sensory characteristics of food is provided, including: establishing a sensory database based on sensory feature data corresponding to the food to be tested; determining the recognition accuracies corresponding to different machine learning models under the sensory characteristics to be tested of the food to be tested based on the sensory database; and determining a target machine learning model for identifying the sensory characteristics to be tested of the food to be tested based on the recognition accuracies corresponding to the different machine learning models.
[0006] According to one aspect of an embodiment of the present application, a model determination device for identifying sensory characteristics of food is provided, including: a data configuration module, configured to establish a sensory database based on sensory feature data corresponding to the food to be tested; a model testing module, configured to determine the recognition accuracies corresponding to different machine learning models under the sensory characteristics to be tested of the food to be tested based on the sensory database; and a model output module, configured to determine a target machine learning model for identifying the sensory characteristics to be tested of the food to be tested based on the recognition accuracies corresponding to the different machine learning models.
[0007] In some embodiments of the present application, based on the aforementioned scheme, the data configuration module is further configured to: obtain a preset score of the food to be tested under the sensory characteristics to be tested; obtain electronic tongue sampling data corresponding to the food to be tested; and use the preset score and the electronic tongue sampling data as the sensory characteristic data.
[0008] In some embodiments of the present application, based on the aforementioned scheme, when the food to be tested includes multiple food samples to be tested, the data configuration module is further configured to: obtain the preset score corresponding to each of the food samples to be tested under the sensory characteristics to be tested; obtain the electronic tongue sampling data corresponding to each of the food samples to be tested; and determine the sensory characteristic data of the food to be tested through the preset score corresponding to each of the food samples to be tested and the electronic tongue sampling data.
[0009] In some embodiments of the present application, based on the aforementioned scheme, the data configuration module is further configured to: obtain a preset standardized processing method corresponding to the sensory characteristic to be tested; perform standardized processing on the electronic tongue sampling data of each of the food samples to be tested based on the preset standardized processing method to obtain the standardized electronic tongue sampling data of each of the food samples to be tested; and use the preset score and the standardized electronic tongue sampling data of each of the food samples to be tested as the sensory characteristic data.
[0010] In some embodiments of the present application, based on the aforementioned scheme, the data configuration module is further configured to: obtain a preset linear dimensionality reduction method corresponding to the sensory characteristic to be tested; perform dimensionality reduction processing on the electronic tongue sampling data of each of the food samples to be tested based on the preset linear dimensionality reduction method to obtain a reduced dimensionality feature vector corresponding to each of the food samples to be tested; and use the preset score, electronic tongue sampling data and the reduced dimensionality feature vector of each of the food samples to be tested as the sensory characteristic data.
[0011] In some embodiments of the present application, based on the aforementioned scheme, the model output module is also configured to: determine a qualified machine learning model that meets the preset screening conditions from the different machine learning models based on the recognition accuracy corresponding to each of the different machine learning models; and determine the target machine learning model through the qualified machine learning model.
[0012] In some embodiments of the present application, based on the aforementioned scheme, when the qualified machine learning model includes multiple machine learning models, the model output module is also configured to: obtain a preset meta-model corresponding to the sensory characteristic to be measured; determine multiple sub-base models through multiple qualified machine learning models; and establish the target machine learning model based on the preset meta-model and the multiple sub-base models.
[0013] According to one aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor of a computer, the computer executes the model determination method for identifying food sensory characteristics as described in the above embodiments.
[0014] According to one aspect of an embodiment of the present application, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements the model determination method for identifying sensory characteristics of food as described in the above embodiments.
[0015] In the technical solution of the embodiment of the present application, a sensory database can be established based on the sensory characteristic data corresponding to the food to be tested, and then the recognition accuracy of different machine learning models under the sensory characteristics to be tested is determined based on the sensory database. Then, based on the recognition accuracy of different machine learning models, a target machine learning model for identifying the sensory characteristics to be tested of the food to be tested is determined, thereby unifying the evaluation criteria by detecting the sensory characteristics of the food with the target machine learning model, thereby effectively avoiding the impact of individual differences among tasters on the final evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The accompanying drawings are incorporated into and constitute a part of the specification, illustrating embodiments consistent with the present application and, together with the specification, serving to explain the principles of the present application. It is obvious that the drawings described below are merely some embodiments of the present application, and a person of ordinary skill in the art can derive other drawings based on these drawings without inventive effort. In the drawings:
[0017] Figure 1 This is a flow chart of a model determination method for food sensory property identification, shown in an exemplary embodiment of the present application;
[0018] Figure 2 yes Figure 1 The flowchart of step S110 in the illustrated embodiment in an exemplary embodiment;
[0019] Figure 3 yes Figure 1 The flowchart of step S130 in the illustrated embodiment in an exemplary embodiment;
[0020] Figure 4 is a block diagram of a model determination device for food sensory property recognition, shown in an exemplary embodiment of the present application;
[0021] Figure 5 It is a structural diagram of an electronic device shown in an exemplary embodiment of the present application. DETAILED DESCRIPTION
[0022] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.
[0023] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner.In the following description, many specific details are provided so as to provide a full understanding of the embodiments of the present application. However, it will be appreciated by those skilled in the art that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps etc. can be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the application.
[0024] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.
[0025] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.
[0026] It should be noted that the term "plurality" used in this document refers to two or more. "And / or" describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. The character " / " generally indicates an "or" relationship between the associated objects.
[0027] The technical solution of the embodiment of the present application proposes a model determination method for food sensory property identification, specifically referring to Figure 1 The method at least includes steps S110 to S130, which are described in detail as follows:
[0028] In step S110, a sensory database is established according to the sensory characteristic data corresponding to the food to be tested.
[0029] In an embodiment of the present application, in order to identify the sensory characteristics of food, a sensory database can be first established based on the sensory characteristic data corresponding to the food to be tested, that is, a sensory database associated with the food to be tested is established, and the sensory characteristic data corresponding to the food to be tested is recorded in the sensory database, wherein the sensory characteristic data represents the sensory system's experience feedback on the food to be tested.
[0030] In order to obtain sensory characteristic data corresponding to the food to be tested, in some embodiments of the present application, electronic tongue sampling data corresponding to the food to be tested may be obtained and used as the sensory characteristic data.
[0031] In order to obtain electronic tongue sampling data corresponding to the food to be tested, an electronic tongue device may be used to sample and analyze the food to be tested, and the sampling signal output by the electronic tongue device may be used as the electronic tongue sampling data. Specifically, the Astree electronic tongue device is used to sample and analyze the food to be tested, that is, the seven sensors on the electronic tongue device and a standard reference electrode are used to detect the food to be tested and output sampling signals, namely, the AHS (Acidity Sensor) sensor, PKS (General Purpose Sensor-PK-type, PK type general sensor) sensor, CTS (Saltiness Sensor) sensor, NMS (Umami Sensor, umami sensor) sensor, CPS (General Purpose Sensor-CP-type, CP type general sensor) sensor, ANS (Sweetness Sensor, sweetness sensor) sensor, SCS (Bitterness Sensor, bitterness sensor) sensor and Ag / AgCl (Silver / Silver Chloride Electrode, silver / silver chloride reference electrode) reference electrode are used to detect the food to be tested and output sampling signals, and then electronic tongue sampling data is generated according to the sampling signals output by each sensor.
[0032] Furthermore, the sampling process can be repeated multiple times to improve the reliability of the electronic tongue sampling data. Furthermore, to enhance the sampling sensitivity of the electronic tongue device, when the food being tested is a liquid, a dilution test can be performed to determine the dilution factor corresponding to the food being tested. After diluting the food according to the dilution factor, the electronic tongue device can be used to sample and analyze the food being tested, ensuring that the sensor of the electronic tongue device is within a reasonable response range, thereby further improving the reliability and accuracy of the electronic tongue sampling data.
[0033] In some embodiments of the present application, a preset score of the food to be tested under the sensory characteristics to be tested can be obtained. The preset score can be an evaluation of the sensory characteristics to be tested after manually tasting the food to be tested, and then the electronic tongue sampling data corresponding to the food to be tested is obtained. Finally, the preset score and the electronic tongue sampling data are used as sensory characteristic data to enhance the correlation between the sensory characteristic data and the food to be tested.
[0034] Among them, in order to obtain the preset score of the food to be tested under the sensory characteristics to be tested, a number of tasters corresponding to the food to be tested can be invited to taste the food to be tested with the help of their professional senses, and evaluate the preset score of the food to be tested. Specifically, a secret evaluation is adopted, and the sensory characteristics of the selected food to be tested are evaluated with the basic food sample of the food to be tested as a reference. It is expected to carry out multiple rounds, and each round will taste and evaluate the food sample to be tested. At the same time, a rest period is set in the middle of each round to facilitate the tasters to restore sensory sensitivity. In addition, the above-mentioned secret evaluation process can be repeated multiple times to improve the accuracy of the preset score of the food to be tested under the sensory characteristics to be tested.
[0035] In some embodiments of the present application, physical and chemical indicators corresponding to the food to be tested can be obtained and used as sensory characteristic data.
[0036] To obtain the physical and chemical indicators corresponding to the food being tested, testing can be performed according to national standard testing methods for each indicator. For example, when the food being tested is liquor, its corresponding physical and chemical indicators include alcohol content, total acidity, solids, viscosity, etc., and the corresponding national standard testing methods include GB 5009.225-2023, GB 12456-2021, GB / T 10345-2007, etc. Furthermore, the above testing process can also use sensors for auxiliary measurement to improve the accuracy of the obtained physical and chemical indicators.
[0037] In addition, before obtaining the physical and chemical indicators corresponding to the food to be tested, you can also first obtain the various indicators corresponding to the food to be tested under different preset scores, and then determine the correlation between the various indicators and the sensory characteristics to be tested based on the various indicators corresponding to the different preset scores, and then determine the physical and chemical indicators corresponding to the food to be tested based on the correlation between the various indicators and the sensory characteristics to be tested, that is, the indicators among the various indicators of the food to be tested that have a higher correlation with the sensory characteristics to be tested are used as physical and chemical indicators.
[0038] In some embodiments of the present application, a preset score of the food to be tested under the sensory characteristics to be tested can be obtained first, and then the physical and chemical indicators corresponding to the food to be tested can be obtained. Finally, the preset score and physical and chemical indicators are used as sensory characteristic data to enhance the correlation between the sensory characteristic data and the food to be tested.
[0039] In some embodiments of the present application, a preset score of the food to be tested under the sensory characteristics to be tested can be obtained, and then the electronic tongue sampling data corresponding to the food to be tested can be obtained, and then the physical and chemical indicators corresponding to the food to be tested can be obtained. Finally, the preset score, electronic tongue sampling data and physical and chemical indicators are used as sensory characteristic data to further enhance the correlation between the sensory characteristic data and the food to be tested.
[0040] In step S120, the recognition accuracy corresponding to different machine learning models under the sensory characteristics of the food to be tested is determined based on the sensory database.
[0041] In the embodiment of the present application, after the sensory database is established, the recognition accuracy of different machine learning models corresponding to the sensory characteristics of the food to be tested can be determined based on the sensory database.
[0042] Among them, machine learning models include but are not limited to random forests, gradient boosting, support vector regression, ensemble learning, elastic networks and other models with prediction or classification capabilities.
[0043] The method of determining the recognition accuracy of different machine learning models under the sensory characteristics of the food to be tested based on the sensory database can be flexibly set as needed. In one example, the sensory feature data recorded in the sensory database can be input into different machine learning models respectively, so that different machine learning models output the predicted scores of the food to be tested under the sensory characteristics to be tested based on the sensory feature data, and then obtain the preset scores of the food to be tested under the sensory characteristics to be tested. Finally, the recognition accuracy corresponding to each machine learning model is determined based on the predicted scores and preset scores output by different machine learning models.
[0044] Among them, the method of determining the recognition accuracy corresponding to each machine learning model based on the predicted scores and preset scores output by different machine learning models can adopt the absolute error method, square error method, absolute percentage error method, etc., which is not limited here.
[0045] In another embodiment, the sensory feature data recorded in the sensory database can be input into different machine learning models respectively, so that different machine learning models output the predicted scores corresponding to the food to be tested under the sensory features to be tested based on the sensory feature data, and then the recognition accuracy corresponding to each machine learning model is directly determined based on the predicted scores output by different machine learning models and the preset scores in the sensory feature data.
[0046] In some embodiments of the present application, under the condition that the machine learning models include random forest, gradient boosting, support vector regression, ensemble learning, and elastic net, and the absolute error method and square error method are used to determine the recognition accuracy corresponding to each machine learning model, the code for determining the recognition accuracy corresponding to each machine learning model under the sensory characteristics of the food to be tested based on the sensory database is as follows:
[0047] (1) Importing the basic library
[0048] import pandas as pd
[0049] import numpy as np
[0050] import matplotlib.pyplot as plt
[0051] import seaborn as sns;
[0052] (2) Importing machine learning models
[0053] from sklearn.ensemble import RandomForestRegressor,GradientBoostingRegressor
[0054] from sklearn.svm import SVR
[0055] from sklearn.linear_model import ElasticNet;
[0056] (3) Derivation of recognition accuracy
[0057] from sklearn.metrics import mean_absolute_error,mean_squared_error.
[0058] In step S130, a target machine learning model for identifying the sensory characteristics of the food to be tested is determined based on the recognition accuracies corresponding to different machine learning models.
[0059] In an embodiment of the present application, after determining the recognition accuracy corresponding to different machine learning models under the sensory characteristics to be tested of the food to be tested, the target machine learning model for identifying the sensory characteristics to be tested of the food to be tested can be determined based on the recognition accuracy corresponding to different machine learning models.
[0060] Among them, the method of determining the target machine learning model used to identify the sensory characteristics of the food to be tested based on the recognition accuracy corresponding to different machine learning models can be flexibly set according to needs. In one example, the highest recognition accuracy can be determined from the recognition accuracies corresponding to different machine learning models, and then the machine learning model corresponding to the highest recognition accuracy can be used as the target machine learning model.
[0061] In another example, a qualified machine learning model that meets preset screening conditions can be determined from different machine learning models based on the recognition accuracy corresponding to each of the different machine learning models, and then the qualified machine learning model can be used as the target machine learning model.
[0062] The screening condition can be an accuracy threshold, meaning a machine learning model is considered qualified if its corresponding recognition accuracy reaches the accuracy threshold. Alternatively, the screening condition can be a TOP-k selection, where k is an adjustable constant, meaning a machine learning model is considered qualified if its corresponding recognition accuracy ranks kth or before the kth of all other machine learning models. The specific preset screening conditions can be adjusted based on the tester's needs and are not limited here.
[0063] Through the above implementation, a sensory database can be established based on the sensory characteristic data corresponding to the food to be tested, and then the recognition accuracy of different machine learning models under the sensory characteristics to be tested of the food to be tested can be determined based on the sensory database. Then, based on the recognition accuracy of different machine learning models, a target machine learning model for identifying the sensory characteristics to be tested of the food to be tested can be determined, thereby unifying the evaluation criteria by detecting the sensory characteristics of the food with the target machine learning model, thereby effectively avoiding the impact of individual differences among tasters on the final evaluation results.
[0064] See also Figure 2 , Figure 2 is Figure 1 The flowchart of step S110 in the embodiment shown is in an exemplary embodiment. Figure 2 As shown, when the food to be tested includes multiple food samples to be tested, the process of establishing a sensory database according to the sensory characteristic data corresponding to the food to be tested may include steps S210 to S230, which are described in detail as follows:
[0065] In step S210, a preset score corresponding to the sensory characteristic to be tested of each food sample to be tested is obtained.
[0066] In step S220, the electronic tongue sampling data corresponding to each food sample to be tested is obtained.
[0067] In step S230, the sensory characteristic data of the food to be tested is determined based on the preset score corresponding to each food sample to be tested and the electronic tongue sampling data.
[0068] In an embodiment of the present application, when the food to be tested includes multiple food samples to be tested, where the ingredient ratios corresponding to each food sample to be tested are different, in the process of establishing a sensory database based on the sensory characteristic data corresponding to the food to be tested, the preset score corresponding to each food sample to be tested under the sensory characteristics to be tested can be obtained first, and then the electronic tongue sampling data corresponding to each food sample to be tested can be obtained. Then, the sensory characteristic data of the food to be tested can be determined by the preset score and electronic tongue sampling data corresponding to each food sample to be tested, so as to increase the sample size of the sensory characteristic data, thereby facilitating the determination of a target machine learning model with stronger recognition ability.
[0069] Secondly, to facilitate access to the sensory characteristic data of each food to be tested in the sensory database, the sensory database can store the sensory characteristic data corresponding to each food to be tested in a table format. Specifically, taking the food to be tested as white wine as an example, the corresponding table of its sensory characteristic data in the sensory database is shown in Table 1.
[0070] Table 1 Sensory characteristics data
[0071]
[0072]
[0073] In addition, when the sensory feature data in the sensory database covers multiple food samples to be tested, in some embodiments of the present application, based on the above content, the above-mentioned step S120 can also be adaptively optimized to further improve the accuracy of the recognition accuracy of each machine learning model determined. That is, the process of determining the recognition accuracy corresponding to each of the different machine learning models under the sensory characteristics to be tested of the food to be tested based on the sensory database can be that after the sensory feature data recorded in the sensory database are respectively input into different machine learning models, the different machine learning models are made to output the prediction scores corresponding to the different food samples to be tested under the sensory characteristics to be tested, and then the recognition accuracy corresponding to each machine learning model is determined based on the prediction scores output by the different machine learning models and the preset score of each food sample to be tested.
[0074] Among them, the recognition accuracy of each machine learning model can be determined based on the predicted scores output by different machine learning models and the preset scores of each food sample to be tested. The determination coefficient method, root mean square error method, cross validation R 2 Laws, etc., are no longer restricted here.
[0075] In some embodiments of the present application, when using the root mean square error method, cross validation R 2 Under the condition that the recognition accuracy corresponding to each machine learning model cannot be determined, the code for determining the recognition accuracy corresponding to different machine learning models under the sensory characteristics of the food to be tested based on the sensory database is as follows:
[0076] (1) Importing the basic library
[0077] import pandas as pd
[0078] import numpy as np
[0079] import matplotlib.pyplot as plt
[0080] import seaborn as sns;
[0081] (2) Importing machine learning models
[0082] from sklearn.ensemble import RandomForestRegressor,GradientBoostingRegressor
[0083] from sklearn.svm import SVR
[0084] from sklearn.linear_model import ElasticNet;
[0085] (3) Derivation of recognition accuracy
[0086] from sklearn.metrics import r2_score,mean_squared_error.
[0087] In the above process, the sensory characteristic data of the food to be tested is determined by the preset score and electronic tongue sampling data corresponding to each food sample to be tested. In one example, the preset score and electronic tongue sampling data corresponding to each food sample to be tested can be directly used as the sensory characteristic data of the food to be tested.
[0088] In another example, a preset standardization processing method corresponding to the sensory characteristic to be tested can be obtained first, and then the electronic tongue sampling data of each food sample to be tested can be standardized based on the preset standardization processing method to obtain the standardized electronic tongue sampling data of each food sample to be tested, and then the preset score of each food sample to be tested and the standardized electronic tongue sampling data can be used as sensory characteristic data, thereby eliminating the dimensional differences in the electronic tongue sampling data of each food sample to be tested due to different types.
[0089] The standardization processing methods include, but are not limited to, the Z-Score standardization method, the Min-Max standardization method, etc. The preset standardization processing method corresponding to the sensory characteristic to be measured can be adjusted according to the needs of the tester and is not limited here.
[0090] In addition, after obtaining the standardized electronic tongue sampling data of each food sample to be tested, the missing values and abnormal values in the electronic tongue sampling data can be further processed to ensure the integrity and accuracy of the standardized electronic tongue sampling data.
[0091] In another example, a preset linear dimensionality reduction method corresponding to the sensory characteristics to be tested can be first obtained, and then the electronic tongue sampling data of each food sample to be tested can be subjected to dimensionality reduction processing based on the preset linear dimensionality reduction method to obtain the corresponding reduced dimensionality feature vector of each food sample to be tested, and then the preset score, electronic tongue sampling data and reduced dimensionality feature vector of each food sample to be tested can be used as sensory feature data to further enhance the correlation between the sensory feature data and the food to be tested.
[0092] Among them, the linear dimensionality reduction method includes but is not limited to principal component analysis, linear discriminant analysis, etc. The preset linear dimensionality reduction method corresponding to the sensory characteristic to be tested can be adjusted according to the needs of the tester, and is not limited here; or, it can be determined based on the verification test, that is, based on each linear dimensionality reduction method to be tested, the electronic tongue sampling data of each food sample to be tested is subjected to dimensionality reduction processing, and the dimensionality reduction feature vector corresponding to each food sample to be tested under the different linear dimensionality reduction methods to be tested is obtained. Then, based on the preset score of each food sample to be tested and the dimensionality reduction feature vector corresponding to the different linear dimensionality reduction methods to be tested, the classification accuracy of each linear dimensionality reduction method to be tested is determined. Finally, based on the classification accuracy of each linear dimensionality reduction method to be tested, the preset linear dimensionality reduction method corresponding to the sensory characteristic to be tested is determined. For example, the highest classification accuracy is determined from the classification accuracies corresponding to the different linear dimensionality reduction methods to be tested, and the linear dimensionality reduction method to be tested corresponding to the highest classification accuracy is used as the preset linear dimensionality reduction method.
[0093] Specifically, the classification accuracy of each linear dimensionality reduction method to be tested can be determined based on the preset score of each food sample to be tested and the corresponding dimensionality reduction feature vector under different linear dimensionality reduction methods to be tested. The determination coefficient method, root mean square error method, relative standard deviation method, etc. are not limited here.
[0094] In some embodiments of the present application, when the food to be tested is white wine, if the linear dimensionality reduction method to be tested includes principal component analysis and linear discriminant analysis, the electronic tongue sampling data of each white wine sample can be subjected to dimensionality reduction processing based on principal component analysis and linear discriminant analysis respectively to obtain the corresponding dimensionality reduction feature vectors of each white wine sample under principal component analysis and linear discriminant analysis, and then the classification accuracy of principal component analysis and linear discriminant analysis is determined based on the preset score of each white wine sample and the corresponding dimensionality reduction feature vectors under principal component analysis and linear discriminant analysis.
[0095] If the determination coefficient method and the root mean square error method are used to determine the classification accuracy of the linear dimensionality reduction method to be tested, the higher the determination coefficient of the linear dimensionality reduction method to be tested and the lower the root mean square error, the higher the classification accuracy. Correspondingly, when it is determined that the classification accuracy of the principal component analysis method is characterized by a determination coefficient of 0.175 and a root mean square error of 1.958, and the classification accuracy of the linear discriminant analysis method is characterized by a determination coefficient of 0.937 and a root mean square error of 0.296, it can be concluded that the principal component analysis method can only explain about 17.5% of the sensory score variation for the liquor samples and the classification error is large. Correspondingly, the linear discriminant analysis method can explain about 93.7% of the sensory score variation for the liquor samples and the classification error is low. Based on this, through the classification accuracy of the principal component analysis method and the linear discriminant analysis method, it can be determined that the preset linear dimensionality reduction method corresponding to the liquor under the current sensory characteristics to be tested is the linear discriminant analysis method.
[0096] In addition, in some embodiments of the present application, when the food to be tested is a liquid food type, it is also possible to obtain a preset dilution multiple of the food to be tested corresponding to the sensory characteristics to be tested, and then obtain the electronic tongue sampling data of the food to be tested at the preset dilution multiple. Then, based on the preset linear dimensionality reduction method, the electronic tongue sampling data of each food sample to be tested is subjected to dimensionality reduction processing to obtain the dimensionality reduction feature vector corresponding to each food sample to be tested, thereby improving the dimensionality reduction effect of the preset linear dimensionality reduction method.
[0097] In order to obtain the preset dilution ratio corresponding to the sensory characteristics of the food to be tested, the electronic tongue sampling data corresponding to the food samples to be tested with different preset scores at different dilution ratios can be obtained in advance, and then the electronic tongue sampling data of each food sample to be tested can be reduced in dimension based on the preset linear dimensionality reduction method to obtain the reduced dimension feature vectors of the food to be tested with different preset scores at different dilution ratios. Then, the preset dilution ratio is determined based on the distribution of the reduced dimension feature vectors of the food to be tested with different preset scores at different dilution ratios. The preset dilution ratio represents the dilution ratio with the highest degree of aggregation in the distribution of the food to be tested with different preset scores.
[0098] For example, when the food to be tested is liquor and the sensory characteristic to be tested is mellowness, the electronic tongue sampling data corresponding to liquors with different mellowness scores at different dilution ratios can be obtained in advance, and then the electronic tongue sampling data of each liquor can be reduced in dimension based on the preset linear dimensionality reduction method to obtain the reduced dimension feature vectors of liquors with different mellowness scores at different dilution ratios. Then, the preset dilution ratio is determined based on the distribution of the reduced dimension feature vectors of liquors with different mellowness scores at different dilution ratios.
[0099] In another example, a preset standardization processing method corresponding to the sensory characteristic to be tested can be first obtained, and then the electronic tongue sampling data of each food sample to be tested can be standardized based on the preset standardization processing method to obtain the electronic tongue sampling data of each food sample to be tested after the standardization processing, and then a preset linear dimensionality reduction method corresponding to the sensory characteristic to be tested can be obtained, and then the electronic tongue sampling data of each food sample to be tested after the standardization processing can be reduced in dimension based on the preset linear dimensionality reduction method to obtain the reduced dimensionality feature vector corresponding to each food sample to be tested, and then the preset score, electronic tongue sampling data and reduced dimensionality feature vector of each food sample to be tested can be used as sensory feature data, so as to combine standardization processing and linear dimensionality reduction to further enhance the correlation between the sensory feature data and the food to be tested.
[0100] In some embodiments of the present application, when the food to be tested includes multiple food samples to be tested, and the sensory characteristic data of the food to be tested are physical and chemical indicators, the sensory characteristic data of the food to be tested are physical and chemical indicators and preset scores, or the sensory characteristic data of the food to be tested are electronic tongue sampling data, physical and chemical indicators and preset scores, it can also be processed according to the above records, and will not be repeated here.
[0101] See also Figure 3 , Figure 3 is Figure 1 The flowchart of step S130 in the embodiment shown is in an exemplary embodiment. Figure 3As shown, the process of determining the target machine learning model for identifying the sensory characteristics of the food to be tested based on the recognition accuracies corresponding to different machine learning models may include steps S310 to S320, which are described in detail as follows:
[0102] In step S310, a qualified machine learning model that meets the preset screening conditions is determined from different machine learning models based on the recognition accuracy corresponding to each of the different machine learning models.
[0103] In an embodiment of the present application, after determining the recognition accuracy corresponding to different machine learning models under the sensory characteristics of the food to be tested, a qualified machine learning model that meets the preset screening conditions can be determined from the different machine learning models based on the recognition accuracy corresponding to the different machine learning models.
[0104] Among them, the method of determining a qualified machine learning model that meets the preset screening conditions from different machine learning models based on the recognition accuracy corresponding to each of the different machine learning models can refer to the record in the above-mentioned step S130 and will not be repeated here.
[0105] In step S320, the target machine learning model is determined through the qualified machine learning models.
[0106] In an embodiment of the present application, after determining a qualified machine learning model that meets the preset screening conditions, the target machine learning model can be determined through the qualified machine learning model.
[0107] Specifically, in some embodiments of the present application, when determining a qualified machine learning model from different machine learning models based on the recognition accuracy corresponding to different machine learning models, when only one of the different machine learning models meets the preset screening conditions, that is, the qualified machine learning model includes a single machine learning model, the qualified machine learning model can be directly used as the target machine learning model.
[0108] In some embodiments of the present application, when determining a qualified machine learning model from different machine learning models based on the recognition accuracy corresponding to different machine learning models, there are multiple machine learning models in the different machine learning models that meet the preset screening conditions, that is, the qualified machine learning model includes multiple machine learning models, then a preset meta-model corresponding to the sensory feature to be tested can be obtained, and then multiple sub-base models can be determined through the multiple qualified machine learning models. Finally, a target machine learning model is established based on the preset meta-model and the multiple sub-base models, so as to improve the recognition ability of the established target machine learning model for the sensory characteristics to be tested of the food to be tested by stacking the machine learning models that meet the preset screening conditions.
[0109] The metamodel is used to integrate the output of the base model. Metamodels include, but are not limited to, linear regression, support vector regression, random forest, neural network, etc. The preset metamodel corresponding to the sensory feature to be tested can be flexibly adjusted according to the needs of the tester and is not limited here.
[0110] The base model is used to output the predicted score for the food under test based on the sensory characteristic data input. The process of determining multiple sub-base models from multiple qualified machine learning models is to use each of the multiple qualified machine learning models as a sub-base model.
[0111] The following describes an embodiment of the device of the present application, which can be used to implement the model determination method for food sensory property identification described in the above-mentioned embodiments of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the model determination method for food sensory property identification described above.
[0112] Figure 4 A block diagram of a model determination device 100 for food sensory property recognition according to an embodiment of the present application is shown.
[0113] Reference Figure 4 As shown, according to an embodiment of the present application, a model determination device 100 for identifying sensory characteristics of food includes: a data configuration module 110, configured to establish a sensory database based on sensory feature data corresponding to the food to be tested; a model testing module 120, configured to determine the recognition accuracies corresponding to different machine learning models under the sensory characteristics to be tested of the food to be tested based on the sensory database; and a model output module 130, configured to determine a target machine learning model for identifying the sensory characteristics to be tested of the food to be tested based on the recognition accuracies corresponding to different machine learning models.
[0114] In some embodiments of the present application, based on the aforementioned scheme, the data configuration module 110 is further configured to: obtain a preset score of the food to be tested under the sensory characteristics to be tested; obtain electronic tongue sampling data corresponding to the food to be tested; and use the preset score and the electronic tongue sampling data as sensory feature data.
[0115] In some embodiments of the present application, based on the aforementioned scheme, when the food to be tested includes multiple food samples to be tested, the data configuration module 110 is further configured to: obtain the preset score corresponding to each food sample to be tested under the sensory characteristics to be tested; obtain the electronic tongue sampling data corresponding to each food sample to be tested; and determine the sensory characteristic data of the food to be tested through the preset score and electronic tongue sampling data corresponding to each food sample to be tested.
[0116] In some embodiments of the present application, based on the aforementioned scheme, the data configuration module 110 is further configured to: obtain a preset standardized processing method corresponding to the sensory characteristics to be tested; standardize the electronic tongue sampling data of each food sample to be tested based on the preset standardized processing method to obtain the standardized electronic tongue sampling data of each food sample to be tested; and use the preset score of each food sample to be tested and the standardized electronic tongue sampling data as sensory characteristic data.
[0117] In some embodiments of the present application, based on the aforementioned scheme, the data configuration module 110 is further configured to: obtain a preset linear dimensionality reduction method corresponding to the sensory characteristics to be tested; perform dimensionality reduction processing on the electronic tongue sampling data of each food sample to be tested based on the preset linear dimensionality reduction method to obtain a reduced dimensionality feature vector corresponding to each food sample to be tested; and use the preset score, electronic tongue sampling data and reduced dimensionality feature vector of each food sample to be tested as sensory feature data.
[0118] In some embodiments of the present application, based on the aforementioned scheme, the model output module 130 is also configured to: determine a qualified machine learning model that meets the preset screening conditions from different machine learning models based on the recognition accuracy corresponding to each of the different machine learning models; and determine the target machine learning model through the qualified machine learning model.
[0119] In some embodiments of the present application, based on the aforementioned scheme, when the qualified machine learning model includes multiple machine learning models, the model output module 130 is also configured to: obtain a preset meta-model corresponding to the sensory characteristics to be measured; determine multiple sub-base models through multiple qualified machine learning models; and establish a target machine learning model based on the preset meta-model and multiple sub-base models.
[0120] It should be noted that the model determination device 100 for identifying food sensory characteristics provided in the above embodiment and the model determination method for identifying food sensory characteristics provided in the above embodiment belong to the same concept, and the specific manner in which each module and unit performs operations has been described in detail in the method embodiment and will not be repeated here.
[0121] An embodiment of the present application further provides an electronic device, including a processor and a memory, wherein the memory stores computer-readable instructions, which, when executed by the processor, implement the model determination method for food sensory property identification as described above.
[0122] Figure 4 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown.
[0123] It should be noted that Figure 4The computer system 200 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.
[0124] like Figure 4 As shown, the computer system 200 includes a central processing unit (CPU) 201, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 202 or the program loaded from the storage part 208 to the random access memory (RAM) 203, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in the RAM 203. The CPU 201, ROM 202 and RAM 203 are connected to each other via a bus 204. An input / output (I / O) interface 205 is also connected to the bus 204.
[0125] The following components are connected to the I / O interface 205: an input section 206 including a keyboard, a mouse, and the like; an output section 207 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 208 including a hard disk; and a communication section 209 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 209 performs communication processing via a network such as the Internet. A drive 210 is also connected to the I / O interface 205 as needed. Removable media 211, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 210 as needed, so that computer programs read therefrom can be installed into the storage section 208 as needed.
[0126] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 209, and / or installed from a removable medium 211. When the computer program is executed by the central processing unit (CPU) 201, the various functions defined in the system of the present application are executed.
[0127] It should be noted that the computer-readable medium shown in the embodiments of the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries a computer-readable computer program. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. A computer program embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, or any suitable combination thereof.
[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the above-mentioned module, program segment or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two boxes shown in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, as well as the combination of boxes in the block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.
[0129] The units described in the embodiments of the present application may be implemented in software or hardware, and the units described may also be provided in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.
[0130] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments, or may exist independently without being incorporated into the electronic device. The computer-readable storage medium carries one or more programs, and when the one or more programs are executed by the electronic device, the electronic device implements the method described in the above embodiments.
[0131] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiment of the application, the features and functions of two or more modules or units described above can be concretized in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.
[0132] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiments of the present application.
[0133] Those skilled in the art will readily conceive of other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of this application and include common knowledge or customary techniques in the art that are not disclosed herein.
[0134] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A model determination method for food sensory property identification, characterized in that: The method comprises: Establishing a sensory database based on the sensory characteristic data corresponding to the food to be tested; Determining, based on the sensory database, the recognition accuracy of different machine learning models corresponding to the sensory characteristics of the food to be tested; A target machine learning model for identifying the sensory characteristics of the food to be tested is determined based on the recognition accuracies corresponding to the different machine learning models.
2. The method according to claim 1, characterized in that The sensory database is established according to the sensory characteristic data corresponding to the food to be tested, including: Obtaining a preset score of the food to be tested under the sensory characteristic to be tested; Obtaining electronic tongue sampling data corresponding to the food to be tested; The preset score and the electronic tongue sampling data are used as the sensory feature data.
3. The method according to claim 1, characterized in that The food to be tested includes a plurality of food samples to be tested, and the sensory database is established according to the sensory characteristic data corresponding to the food to be tested, including: Obtaining a preset score corresponding to the sensory characteristic to be tested for each of the food samples to be tested; Obtaining electronic tongue sampling data corresponding to each of the food samples to be tested; The sensory characteristic data of the food to be tested is determined by the preset score corresponding to each of the food samples to be tested and the electronic tongue sampling data.
4. The method according to claim 3, characterized in that The determining of the sensory characteristic data of the food to be tested by using the preset score corresponding to each of the food samples to be tested and the electronic tongue sampling data includes: Obtaining a preset standardized processing method corresponding to the sensory characteristic to be measured; Performing standardization processing on the electronic tongue sampling data of each of the food samples to be tested based on the preset standardization processing method to obtain the electronic tongue sampling data of each of the food samples to be tested after the standardization processing; The preset score of each food sample to be tested and the standardized electronic tongue sampling data are used as the sensory characteristic data.
5. The method according to claim 3, characterized in that The determining of the sensory characteristic data of the food to be tested by using the preset score corresponding to each of the food samples to be tested and the electronic tongue sampling data includes: Obtaining a preset linear dimensionality reduction method corresponding to the sensory characteristic to be measured; Performing dimensionality reduction processing on the electronic tongue sampling data of each of the food samples to be tested based on a preset linear dimensionality reduction method to obtain a dimensionality reduction feature vector corresponding to each of the food samples to be tested; The preset score, electronic tongue sampling data and the dimension reduction feature vector of each food sample to be tested are used as the sensory feature data.
6. The method according to claim 1, characterized in that The determining of a target machine learning model for identifying the sensory property of the food to be tested based on the recognition accuracies corresponding to the different machine learning models includes: Determining a qualified machine learning model that meets a preset screening condition from the different machine learning models based on the recognition accuracy corresponding to each of the different machine learning models; The target machine learning model is determined by the qualified machine learning model.
7. The method according to claim 6, characterized in that The qualified machine learning model includes a plurality of machine learning models, and determining the target machine learning model through the qualified machine learning models includes: Obtaining a preset meta-model corresponding to the sensory characteristic to be measured; Determine a plurality of sub-base models using a plurality of the qualified machine learning models; The target machine learning model is established based on the preset meta-model and the multiple sub-base models.
8. A model determination device for food sensory property identification, characterized in that: include: A data configuration module configured to establish a sensory database based on sensory characteristic data corresponding to the food to be tested; a model testing module configured to determine, based on the sensory database, the recognition accuracy of different machine learning models corresponding to the sensory characteristics of the food to be tested; The model output module is configured to determine a target machine learning model for identifying the sensory characteristics of the food to be tested based on the recognition accuracies corresponding to the different machine learning models.
9. A computer-readable storage medium, characterized in that Computer-readable instructions are stored thereon, and when the computer-readable instructions are executed by a processor of a computer, the computer is caused to execute the model determination method for identifying food sensory characteristics according to any one of claims 1 to 7.
10. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the model determination method for food sensory property identification as described in any one of claims 1 to 7.