Traditional Chinese medicinal material efficacy evaluation method and device based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based traditional Chinese medicinal material efficacy evaluation method and device, which can fully mine internal associations and modes in data based on artificial intelligent processing of large-scale multi-dimensional data, and finally improve the accuracy and stability of traditional Chinese medicinal material efficacy evaluation. The traditional Chinese medicinal material efficacy evaluation method comprises the following steps: performing data acquisition and data annotation in a traditional Chinese medicinal material production base, and storing the acquired data attributes in a structured JSON format; expanding the data by adopting a multiple linear interpolation algorithm based on a feature coupling degree, and generating a new data point by evaluating the coupling degree between input data features; performing feature extraction on the expanded data through a six-layer full-connection neural network; carrying out dimension reduction processing on the extracted features by adopting an auto-encoder based on reverse reconstruction; classifying the data after dimension reduction by using a support vector machine algorithm based on quantum state mapping; and processing a new traditional Chinese medicinal material sample by using the trained model so as to evaluate the efficacy of the traditional Chinese medicinal material.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of data processing technology, and in particular to a method and device for evaluating the efficacy of traditional Chinese medicine based on artificial intelligence. Background Art

[0002] The efficacy evaluation of traditional Chinese medicinal materials mainly relies on experience and experiments. These methods are usually time-consuming and costly, and the results are greatly affected by subjective factors, making it difficult to achieve large-scale, standardized efficacy evaluation.

[0003] With the development of artificial intelligence technology, especially the progress of machine learning and deep learning, the use of these technologies for drug efficacy evaluation provides new possibilities. In the production and research of traditional Chinese medicine, there are many types of medicinal materials, and the composition and efficacy of each medicinal material are complex and varied, making it particularly important to accurately evaluate the efficacy of each medicinal material.

[0004] However, existing technologies have certain limitations in data processing and feature extraction. Especially when processing large-scale multi-dimensional data, it is often difficult to fully explore the inherent correlations and patterns in the data, resulting in inaccurate or unstable evaluation results. Summary of the invention

[0005] In order to overcome the above technical defects, the present invention provides a method and device for evaluating the efficacy of Chinese herbal medicines based on artificial intelligence, which are as follows:

[0006] In a first aspect, an embodiment of the present invention provides a method for evaluating the efficacy of Chinese herbal medicines based on artificial intelligence, comprising:

[0007] S1. Data collection and annotation: Data collection and annotation are performed at the Chinese medicinal materials production base. The collected data attributes include the growth environment of the medicinal materials and the results of component analysis. The collected data attributes are stored in a structured JSON format.

[0008] S2, data expansion: The data is expanded using a multilinear interpolation algorithm based on feature coupling, and new data points are generated by evaluating the coupling between input data features;

[0009] S3, feature extraction model training: extract features from the expanded data through a six-layer fully connected neural network;

[0010] S4, feature dimension reduction model training: The extracted features are reduced in dimension using an autoencoder based on inverse reconstruction, and the encoding quality and decoding accuracy are optimized by minimizing the dependencies between different features;

[0011] S5, classifier model training: Use the support vector machine algorithm based on quantum state mapping to classify the reduced-dimensional data;

[0012] S6. Evaluation of the efficacy of Chinese medicinal materials: Use the trained model to process new Chinese medicinal materials samples to evaluate their efficacy.

[0013] Optionally, in a possible implementation manner of the first aspect, the specific steps of the data expansion include:

[0014] S201, performing feature coupling analysis on the collected raw data, identifying the interdependence between the features, and obtaining coupling analysis results;

[0015] S202, according to the coupling degree analysis result, selecting features with high coupling degree as selected coupling features for subsequent data generation;

[0016] S203, during the data generation process, setting adaptive oscillation parameters to control the amplitude and direction of interpolation;

[0017] S204, performing multiple linear interpolation using the selected coupling features and the set adaptive oscillation parameters to generate a new data set, wherein the interpolation function is dynamically adjusted according to the coupling degree of the features to generate data that conforms to the actual distribution;

[0018] S205, evaluating the quality of the new data set by similarity with the statistical distribution of the original data, and adjusting the adaptive oscillation parameters according to the evaluation results to optimize the data generation process;

[0019] S206 , after the adaptive oscillation parameter is adjusted, the data generation process is repeatedly iterated until a preset number of iterations is met.

[0020] Optionally, in a possible implementation manner of the first aspect, the specific steps of training the feature dimensionality reduction model include:

[0021] S301, initializing the parameters of the neural network in the feature extraction model, the parameters including the weight and bias parameters of the neural network;

[0022] S302, after the parameters of the neural network are initialized or optimized and updated, when the input data passes through each layer of the neural network, a linear transformation and a nonlinear activation function are performed to output a corresponding feature representation;

[0023] S303, using an exploratory gradient descent update strategy to predict gradient changes, and adjusting the step size and direction of the gradient update according to the prediction results to optimize and update the parameters of the neural network;

[0024] S304: Repeat the above steps S301-S303 until the preset stop iteration condition is met, and the model training is completed.

[0025] Optionally, in a possible implementation manner of the first aspect, the specific steps of training the feature dimensionality reduction model include:

[0026] S401, initializing the parameters of the autoencoder algorithm, the parameters including the weight and bias parameters of the autoencoder;

[0027] S402, in the forward propagation process, the input data is first converted into a representation of a low-dimensional space by the encoder, and then the representation of the low-dimensional space is reconstructed into output data with the same dimension as the original data by the decoder;

[0028] S403, using the inverse reconstruction loss function to calculate the inverse reconstruction loss of the autoencoder;

[0029] S404, using a gradient descent algorithm to update the weights and bias parameters of the encoder and decoder according to the gradient calculated by the inverse reconstruction loss function;

[0030] S405: Repeat the above steps S401-S404 until the preset stop iteration condition is met, and the model training is completed.

[0031] Optionally, in a possible implementation manner of the first aspect, the specific steps of training the classifier model include:

[0032] S501, converting the data after feature dimension reduction into quantum state representation, each data point is encoded as a quantum bit sequence, and its probability amplitude is used to represent the characteristics of the data;

[0033] S502, using a quantum state mapping kernel function to calculate the similarity between two data points through the inner product of the quantum state;

[0034] S503, optimizing the classifier model using an optimization objective function based on density estimation;

[0035] S504: Based on the above optimization objective function, the weights and bias parameters of the classifier model are optimized through quantum state mapping and quantum computing to achieve the optimal classification effect and complete the classifier model training.

[0036] In a second aspect, an embodiment of the present invention provides a Chinese herbal medicine efficacy evaluation device based on artificial intelligence, comprising:

[0037] at least one processor; and a memory in communication with the at least one processor; wherein:

[0038] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described in any one of the first aspect and its implementation methods.

[0039] The above technical scheme is adopted in the embodiment of the present invention to achieve the following technical effects: based on artificial intelligence, the model training algorithm and its model are improved from the aspects of data expansion, feature extraction, feature dimension reduction and classifier, so that the intrinsic correlation and patterns in the data can be fully mined when processing large-scale multi-dimensional data, and finally the accuracy and stability of the efficacy evaluation of traditional Chinese medicine can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings exemplarily illustrate the embodiments and constitute a part of the specification, and together with the text description of the specification, are used to explain the exemplary implementation of the embodiments. The embodiments shown are for illustrative purposes only and do not limit the scope of the claims. In all drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0041] Figure 1 Schematic diagram of an evaluation process of a method for evaluating the efficacy of Chinese herbal medicines based on artificial intelligence in an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of a data expansion process for performing data expansion based on a multilinear interpolation algorithm of feature coupling degree in an embodiment of the present invention;

[0043] Figure 3 A schematic diagram of a training process for training a feature extraction model based on a neural network algorithm for exploratory gradient descent in an embodiment of the present invention;

[0044] Figure 4 A schematic diagram of a training process for training a feature dimensionality reduction model using an autoencoder algorithm based on inverse reconstruction in an embodiment of the present invention;

[0045] Figure 5 A schematic diagram of a training process for training a classifier model using a support vector machine algorithm based on quantum state mapping in an embodiment of the present invention;

[0046] Figure 6 Schematic diagram of the hardware structure of the Chinese medicinal material efficacy evaluation device in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0048] It should be noted that the descriptions involving "first", "second", etc. in the embodiments of the present invention are only for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0049] In the description of the present invention, it should be understood that the numerical labels before the steps do not identify the order of executing the steps, but are only used to facilitate the description of the present invention and distinguish each step, and therefore cannot be understood as a limitation of the present invention.

[0050] Embodiment 1

[0051] See also Figure 1 In the embodiment of the present invention, the method for evaluating the efficacy of Chinese medicinal materials based on artificial intelligence includes:

[0052] S1. Data collection and annotation: Data is collected and annotated at the Chinese medicinal materials production base. The collected data attributes include the growth environment of the medicinal materials and the results of component analysis. The collected data attributes are stored in a structured JSON format.

[0053] Specifically, the attributes of the data include: a1 is the type of medicinal material (such as ginseng, astragalus), R a2 is the collection location, R a3 is the acquisition time, R a4 is the maturity of medicinal materials, R a5 is the water content, R a6 is the content of main chemical components, R a7 is color, R a8 For smell, R a9 For texture, R a10 The recovery rate of patients.

[0054] It should be noted that this embodiment is only used to illustrate one data format and type of the present invention. In actual applications, the attributes of data are usually more than 10 attributes, and the number of attributes of data may reach dozens or even hundreds.

[0055] Furthermore, in the present invention, the collected data are annotated manually. Specifically, the annotated categories may be the following four categories: “very effective”, “effective”, “general” and “ineffective”.

[0056] S2. Data expansion: The data is expanded using a multilinear interpolation algorithm based on feature coupling, and new data points are generated by evaluating the coupling between input data features.

[0057] In the task of the present invention, the collection, acquisition, labeling and preprocessing of training data are time-consuming and labor-intensive, and insufficient training samples can easily lead to poor generalization ability of the model, while affecting the accuracy of the model. The present invention adopts a multilinear interpolation algorithm based on feature coupling to generate new data points by evaluating the coupling between input data features. The analysis of the coupling enables the algorithm to identify and utilize the intrinsic relationship in the Chinese medicinal material data, thereby generating realistic data samples that conform to the actual distribution.

[0058] S3. Feature extraction model training: Feature extraction of the expanded data is performed through a six-layer fully connected neural network.

[0059] Specifically, a 6-layer fully connected neural network, including four hidden layers, has the following structure:

[0060] a-Input layer: receives and transmits sample data after sample expansion;

[0061] b-Hidden layers: The network contains four hidden layers, each consisting of a certain number of neurons;

[0062] b1-first hidden layer: has 128 neurons, used for preliminary feature extraction;

[0063] b2-second hidden layer: has 256 neurons to further abstract features;

[0064] b3-the third hidden layer: has 512 neurons, further deepening feature extraction;

[0065] b4-the fourth hidden layer: has 1024 neurons, further deepening feature extraction;

[0066] b5 - output layer: has 512 neurons.

[0067] S4. Feature dimension reduction model training: The extracted features are reduced in dimension using an autoencoder based on inverse reconstruction, and the encoding quality and decoding accuracy are optimized by minimizing the dependencies between different features.

[0068] The data after feature extraction is input into the feature dimensionality reduction model to train the feature dimensionality reduction model. The present invention adopts an autoencoder algorithm based on inverse reconstruction as the feature dimensionality reduction model. The autoencoder includes two parts: an encoder and a decoder. The encoder is responsible for compressing the input data into a low-dimensional representation, and the decoder reconstructs the data from this low-dimensional representation. In order to enhance the adaptability of the feature dimensionality reduction of the model, the present invention adopts an inverse reconstruction strategy to process the high-dimensional data after input feature extraction, so as to minimize the dependency between different features, thereby improving the quality of encoding and the accuracy of decoding.

[0069] S5. Classifier model training: Use the support vector machine algorithm based on quantum state mapping to classify the reduced-dimensional data. The classification results are "very effective", "effective", "general" and "ineffective"

[0070] The reduced-dimensional data is input into a classifier to train the classifier model. The present invention adopts a support vector machine algorithm based on quantum state mapping as a classification algorithm, adopts a quantum state mapping mechanism in the kernel function of the support vector machine, and uses quantum bits to represent data points, thereby effectively capturing the high-dimensional nonlinear characteristics of the data, so that the model has higher accuracy and stability when processing complex Chinese medicinal materials data.

[0071] S6. Evaluation of the efficacy of Chinese medicinal materials: Use the trained model to process new Chinese medicinal materials samples to evaluate their efficacy.

[0072] In one embodiment, the collected raw data is input into the trained feature extraction and feature dimension reduction model for feature processing, and further, the processed features are input into the classifier model for classifier training to obtain classification results. In this embodiment, the classification categories include "very effective", "effective", "general", and "ineffective", a total of 4 categories.

[0073] The following technical effects can be achieved through the technical solutions in the embodiments of the present invention:

[0074] First, the present invention performs data expansion through a multilinear interpolation algorithm based on feature coupling, uses mutual information to evaluate the coupling between input data features, and generates realistic data samples. On the basis of the original data set, it enhances the richness and representativeness of the samples and avoids the problem of insufficient model training due to insufficient samples.

[0075] Second, in terms of feature extraction, the present invention adopts a fully connected neural network, and a training mechanism based on exploratory gradient descent is embedded in the network. By predicting the changing trend of the gradient and adjusting the learning step size and direction, the problem of gradient disappearance or explosion is effectively avoided, thereby improving the stability of the model and the training efficiency.

[0076] Third, the feature dimensionality reduction of the present invention adopts an autoencoder algorithm based on inverse reconstruction, which minimizes the dependency between different features when compressing data, optimizes the quality of encoding and the accuracy of decoding, and improves data processing efficiency and the prediction accuracy of the model.

[0077] Fourth, the classifier model training of the present invention adopts a support vector machine algorithm based on quantum state mapping, which uses quantum bits to represent the high-dimensional nonlinear characteristics of data points, thereby enhancing the model's ability to process complex data and classification accuracy.

[0078] Based on artificial intelligence, the model training algorithm and its model are improved in terms of data expansion, feature extraction, feature dimensionality reduction and classifier. When processing large-scale multi-dimensional data, it can fully explore the inherent correlations and patterns in the data, and ultimately improve the accuracy and stability of the efficacy evaluation of traditional Chinese medicine.

[0079] Embodiment 2

[0080] The data expansion in the embodiment of the present invention is performed based on a multilinear interpolation algorithm of feature coupling degree, such as Figure 2 As shown in Figure 2, the specific data expansion process is as follows:

[0081] S201, performing feature coupling analysis on the collected raw data, identifying the interdependence between the features, and obtaining coupling analysis results.

[0082] The feature coupling degree of the original data set is analyzed to identify the interdependence between the features. Specifically, the mutual information I(X c ; Y c ) is used to measure the interdependence between two features and, and the calculation method is expressed as:

[0083]

[0084] Where, X c Represents the first feature set of the original data, Y c Represents the second feature set of the original data, x c Represents the first feature of the original data, y c Represents the second feature of the original data, p(x c ,y c ) represents the joint probability density function, p(x c ) indicates X c The marginal probability density function, p(y c ) indicates Y c The marginal probability density function of .

[0085] Furthermore, the calculation method of the joint probability density function and the marginal probability density function is expressed as:

[0086]

[0087]

[0088]

[0089] Where Kd() is the kernel function, usually a Gaussian kernel, nd is the number of sample points, and x ci and ci Represents the X of the i-th data point in the sample c and Y c Eigenvalue.

[0090] S202: According to the coupling degree analysis result, select features with high coupling degree as selected coupling features for subsequent data generation.

[0091] According to the coupling analysis results, feature pairs with high coupling are selected for subsequent data generation. These feature pairs are more likely to contain valuable information for model training. The conditions for data generation are expressed as:

[0092]

[0093] In the formula, C(X c , Y c ) is the data generation indicator function, indicating whether to select the feature pair (X c , Y c ) for data expansion, θ c is the preset data generation threshold.

[0094] S203 . During the data generation process, set adaptive oscillation parameters to control the amplitude and direction of interpolation.

[0095] Set the adaptive oscillation parameters, which control the amplitude and direction of interpolation during data generation to simulate the dynamic relationship and changes between features, expressed as:

[0096] α c =λ c ·exp(-γ c ·I(X c ; Y c ));

[0097] In the formula, λ c is the oscillation baseline strength; γ c is the attenuation coefficient; α c is the adaptive oscillation parameter.

[0098] Furthermore, the calculation method of the reference oscillation intensity is expressed as:

[0099]

[0100] In the formula, σ c is the adjustment coefficient, var(X c ) and var(Y c ) are X c and Y c The variance of .

[0101] And, the calculation method of the attenuation coefficient is expressed as:

[0102]

[0103] In the formula, μ c Yes X c and Y c The mean coupling degree between features, ∈ c is a small positive number that prevents division by zero. Preferably, ∈ c Set to 0.01.

[0104] S204 , using the selected coupling features and the set adaptive oscillation parameters, perform multiple linear interpolation to generate a new data set, wherein the interpolation function is dynamically adjusted according to the coupling degree of the features to generate data that conforms to the actual distribution.

[0105] With the selected coupling feature and the set oscillation parameter α c , perform multiple linear interpolation to generate new data points, and the interpolation function is dynamically adjusted according to the coupling degree of the features to generate high-quality data that conforms to the actual distribution, expressed as:

[0106] Z c =X c +α c ·(Y c -X c )·C(X c , Y c );

[0107] In the formula, Z c is the generated new data point, C(X c , Y c ) is the interpolation adjustment factor.

[0108] Furthermore, the calculation method of the interpolation adjustment factor is expressed as:

[0109]

[0110] Wherein, kd is a sensitivity adjustment factor. Preferably, kd is set to 3.

[0111] S205 , evaluating the quality of the new data set by comparing its similarity to the statistical distribution of the original data, and adjusting the adaptive oscillation parameters according to the evaluation results to optimize the data generation process.

[0112] The quality of the newly generated dataset will be evaluated by the similarity S with the statistical distribution of the original data. c To evaluate, and adjust the oscillation parameters according to the evaluation results, optimize the data generation process, expressed as:

[0113]

[0114]

[0115] In the formula, express and The Kullback-Leibler divergence between two distributions, Z c The distribution of For X c The distribution of c is the sensitivity parameter, β c Adjust the rate at which parameters are updated.

[0116] Further, and The Kullback-Leibler divergence between two distributions is calculated as:

[0117]

[0118]

[0119] Where KL(||) represents the Kullback-Leibler divergence, Md represents and The mean of two distributions.

[0120] S206 , after the adaptive oscillation parameter is adjusted, the data generation process is repeatedly iterated until a preset number of iterations is met.

[0121] During the repeated iterations, the parameter α c The update is repeated until the preset number of iterations N is met. c After the iteration is completed, a sample generated after interpolation is obtained. In one embodiment, when the number of samples reaches a preset total number, it indicates that the sample expansion is completed.

[0122] Embodiment 3

[0123] After completing the data expansion of the original data, the expanded data is input into the feature extraction model for model training, as described above: In the present invention, a 6-layer fully connected neural network is used for feature extraction.

[0124] When using neural networks for feature extraction in some existing solutions, in some neural network structures, problems such as gradient vanishing, gradient explosion, or falling into local optimal solutions may occur, affecting the stability of training and the performance of the model. The present invention adopts a neural network based on exploratory gradient descent as a feature extraction model. In the traditional neural network training process, the exploratory gradient descent method is used to not only consider the current gradient information, but also adopt a forward exploration mechanism to predict the possible change trend of the gradient before the gradient is updated, and adjust the learning step size and direction to achieve faster convergence speed and better local minimum, thereby enhancing the generalization ability of the feature.

[0125] Specifically, if Figure 3 As shown, the training process of the neural network algorithm based on exploratory gradient descent in the feature extraction model of this application is as follows:

[0126] S301, initializing the parameters of the neural network in the feature extraction model, the parameters including the weight and bias parameters of the neural network.

[0127] Initialize the parameters of the neural network, including the weight and bias parameters of the neural network. In one embodiment, the initialization method is expressed as:

[0128]

[0129]

[0130] Where W p，i is the weight matrix of the i-th layer of the neural network; b p，i is the bias vector of the i-th layer of the neural network; σ p is the scaling factor of the neural network weights, β p is the scaling factor of the bias of the neural network; represents a normal distribution with a mean of 0 and a standard deviation of 1; ∈ p is a small noise. Preferably, ∈ p is a noise vector whose elements are all 0.001.

[0131] S302: After the parameters of the neural network are initialized or optimized and updated, when the input data passes through each layer of the neural network, linear transformation and nonlinear activation function processing are performed to output corresponding feature representations.

[0132] The input data passes through each layer of the neural network, and each layer is processed by linear transformation and nonlinear activation function, and finally the feature representation of the output layer is obtained. The output calculation method of each layer is:

[0133] a p，i =Rez(W p，i ·a p，i-1 +b p，i )

[0134] In the formula, a p，i represents the activation output of the i-th layer of the neural network; a p，i-1 is the output of the i-1th layer of the neural network; Rez() is the adaptive ReLU activation function.

[0135] Optionally, in some implementations, let the input of the adaptive ReLU activation function be x zsy , the calculation method of the adaptive ReLU activation function is expressed as:

[0136] Rez(x zsy )=γ p max(0, x zsy )+ξ p min(0, x zsy );

[0137] In the formula, γ p and p are the adaptive coefficients of the positive and negative parts respectively, max() is the maximum value function, and min() is the minimum value function. Preferably, γ p Set to 0.3, ξ p Set to 0.7.

[0138] S303, using an exploratory gradient descent update strategy to predict gradient changes, and adjusting the step size and direction of the gradient update according to the prediction results to optimize and update the parameters of the neural network.

[0139] Adopting the exploratory gradient descent update strategy, before each update of the neural network parameters, the possible gradient changes are predicted, and the step size and direction of the gradient update are adjusted according to the prediction results to optimize the parameter adjustment during the training process and avoid falling into the local optimum too early. The calculation method is expressed as:

[0140]

[0141] In the formula, ΔW p，i and Δb p，i are the update amounts of the i-th layer weight and i-th layer bias of the neural network respectively; η p is the learning rate of the neural network; is the loss function with respect to parameter W p，i The gradient of is the loss function with respect to parameter b p，i The gradient of is the loss function of the neural network; α p is the exploration intensity coefficient; is the prediction function of the gradient c. Preferably, η p Set to 0.03, Cross entropy loss is used.

[0142] Optionally, in some implementations, the exploratory gradient prediction function uses historical gradient data to perform linear regression prediction, and the calculation method is expressed as:

[0143]

[0144] In the formula, Represents the loss function at the t-1th iteration with respect to the parameter W p，i The gradient of t is the attenuation factor; α p is the exploration intensity coefficient; Tgz is the total number of iterations. Preferably, β t Set to 0.98, alpha p Set to 2.

[0145] Furthermore, the weight and bias parameters of the neural network are updated by using the error back propagation method.

[0146] S304: Repeat the above steps S301-S303 until the preset stop iteration condition is met, and the model training is completed.

[0147] The preset condition for stopping the iteration is that a preset maximum number of iterations is reached. For example, the preset maximum number of iterations is set to 1000 times.

[0148] Embodiment 4

[0149] like Figure 4 As shown, the training process of the autoencoder algorithm based on inverse reconstruction in the feature dimensionality reduction model of the present invention specifically includes:

[0150] S401, initializing the parameters of the autoencoder algorithm, which include the weight and bias parameters of the autoencoder.

[0151] Initialize the parameters of the autoencoder algorithm, including the weight and bias parameters of the autoencoder. In some embodiments, the initialization method is expressed as:

[0152]

[0153] b r，init =0;

[0154] Where Wr，init is the initial weight matrix of the autoencoder; b r，init is the initial bias of the autoencoder; n r，in and n r，out Represent the number of neurons in the input and output layers respectively; U(-1, 1) represents a random number uniformly distributed in the interval [-1, 1].

[0155] S402. In the forward propagation process, the input data is first converted into a representation in a low-dimensional space by the encoder, and then the representation in the low-dimensional space is reconstructed into output data with the same dimension as the original data by the decoder.

[0156] The input feature-extracted data is converted into a low-dimensional representation through the encoder, and the low-dimensional representation is reconstructed into an output with the same dimension as the original data through the decoder. In the forward propagation process, the encoder output calculation method is expressed as:

[0157] z r =Sig(W r，enc ·x r +b r，enc );

[0158] And, the reconstructed output of the decoder is:

[0159]

[0160] In the formula, z r is the low-dimensional feature representation of the code; Sig() is the Sigmoid activation function; W r，enc and W r，dec are the weights of the encoder and decoder respectively; b r，enc and b r，dec are the biases of the encoder and decoder respectively; x r is the input data; is the output of the reconstruction.

[0161] S403: Calculate the inverse reconstruction loss of the autoencoder using the inverse reconstruction loss function.

[0162] Calculate the inverse reconstruction loss of the autoencoder and define the calculation method of the inverse reconstruction loss function as:

[0163]

[0164] Where, L r is the inverse reconstruction loss function; Ntr is the number of samples input in the current batch; |||| represents the L2 norm; λ r is the regularization parameter; D KL (P r ||Q r ) is the KL divergence.

[0165] In one embodiment, KL divergence measures the original data distribution P r and reconstruct the data distribution Q r The difference between them is used to ensure that the key features of the data are not lost during the dimensionality reduction process. The calculation method is expressed as:

[0166]

[0167] In the formula, p r，i and q r，i are the probability distributions of the original data and the reconstructed data in the i-th dimension, respectively, and Mtr is the feature dimension before feature dimensionality reduction.

[0168] S404: According to the gradient calculated by the inverse reconstruction loss function, the weight and bias parameters of the encoder and decoder are updated using a gradient descent algorithm.

[0169] According to the gradient calculated by the loss function, the weights and biases of the encoder and decoder are updated using the gradient descent algorithm. The update method is expressed as:

[0170]

[0171]

[0172]

[0173]

[0174] Where η r is the learning rate; is the updated encoder weight; is the updated encoder bias; is the updated decoder weight; is the updated decoder bias.

[0175] Furthermore, the gradient For example, the calculation method is expressed as:

[0176]

[0177] In the formula, represents the partial derivative of the loss function with respect to the reconstructed output, is the partial derivative during decoding, is the partial derivative during encoding.

[0178] S405: Repeat the above steps S401-S404 until the preset stop iteration condition is met, and the model training is completed.

[0179] The preset condition for stopping iteration is reaching a preset maximum number of iterations. Preferably, the preset maximum number of iterations is set to 1200 times.

[0180] Embodiment 5

[0181] like Figure 5 As shown, the training process of the support vector machine algorithm based on quantum state mapping in the classifier model of the present invention is as follows:

[0182] S501. Convert the data after feature dimensionality reduction into quantum state representation. Each data point is encoded as a quantum bit sequence, and its probability amplitude is used to represent the characteristics of the data.

[0183] The data after feature dimensionality reduction is converted into quantum state representation. Each data point is encoded as a quantum bit sequence, and its probability amplitude is used to represent the characteristics of the data. Specifically, for each data point after feature dimensionality reduction, Represent it as a quantum state The present invention adopts quantum coding, each feature x ui is converted into the phase and amplitude of quantum bits, expressed as:

[0184]

[0185] In the formula, x ui is the i-th eigenvalue, θ ui is the eigenvalue conversion angle; φ ui is the phase encoding; |0> and |1> represent the ground state of the quantum bit; |> represents the right vector; i * is an imaginary unit; is element-wise multiplication, and nu is the number of sample eigenvalues.

[0186] In some embodiments, the phase encoding is calculated as:

[0187]

[0188] And, the calculation method of the eigenvalue conversion angle is expressed as:

[0189]

[0190] In the formula, min(X u ) and max(X u ) are the features X u The minimum and maximum values ​​among all samples.

[0191] S502: Use the quantum state mapping kernel function to calculate the similarity between two data points through the inner product of the quantum state.

[0192] Construct a quantum state mapping kernel function, wherein the quantum state mapping kernel function calculates the similarity between two data points through the inner product of the quantum state. In one embodiment, the quantum kernel function Through the inner product calculation of the quantum state, it is expressed as:

[0193]

[0194] In the formula, To distinguish from data points Another data point of , <| indicates the left arrow, It is a data point The quantum state of .

[0195] S503: Optimize the classifier model using an optimization objective function based on density estimation.

[0196] The optimization objective function based on density estimation is defined, which not only considers the maximization of classification interval, but also uses the regularization term based on density estimation to improve the performance of the model in data sparse areas. The calculation method of the optimization objective function is expressed as:

[0197]

[0198] In the formula, Minimize and b u parameter; is the weight vector; b u is the bias term; ui is the slack variable; C is the regularization parameter; λ u is the weight of density-dependent regularization; R u () is a regularization term. Preferably, λ u Set to 0.3.

[0199] Furthermore, the constraints of the above objective function are:

[0200]

[0201] In the formula, y ui () represents the support vector machine classification function.

[0202] Furthermore, the calculation method of the regularization term is expressed as:

[0203]

[0204] In the formula, Represents the i-th eigenvalue of the data point The density estimate, K h (,) is the kernel function with bandwidth h, x uj is the jth eigenvalue.

[0205] S504: Based on the above optimization objective function, the weights and bias parameters of the classifier model are optimized through quantum state mapping and quantum computing to achieve the optimal classification effect and complete the classifier model training.

[0206] Embodiment 6

[0207] Figure 6 The schematic diagram of the hardware architecture of a Chinese medicinal material efficacy evaluation device 10000 suitable for implementing a Chinese medicinal material efficacy evaluation method based on artificial intelligence is shown schematically. In some embodiments, the Chinese medicinal material efficacy evaluation device 10000 may be a terminal device such as a smart phone, a wearable device, a tablet computer, a personal computer, a vehicle-mounted terminal, a game console, a virtual device, a workbench, a digital assistant, a set-top box, a robot, etc. In other embodiments, the Chinese medicinal material efficacy evaluation device 10000 may be a rack server, a blade server, a tower server, or a cabinet server (including an independent server, or a server cluster composed of multiple servers), etc. Figure 6 As shown, the Chinese herbal medicine efficacy evaluation device 10000 includes but is not limited to: a memory 10010, a processor 10020, and a network interface 10030 which can communicate with each other through a system bus. Among them:

[0208] The memory 10010 includes at least one type of computer-readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 10010 can be an internal storage module of the Chinese medicinal material efficacy evaluation device 10000, such as a hard disk or memory of the Chinese medicinal material efficacy evaluation device 10000. In other embodiments, the memory 10010 can also be an external storage device of the Chinese medicinal material efficacy evaluation device 10000, such as a plug-in hard disk equipped on the Chinese medicinal material efficacy evaluation device 10000, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. Of course, the memory 10010 may also include both the internal storage module of the Chinese herbal medicine efficacy evaluation device 10000 and its external storage device. In this embodiment, the memory 10010 is generally used to store the operating system and various application software installed in the Chinese herbal medicine efficacy evaluation device 10000, such as the program code of the Chinese herbal medicine efficacy evaluation method based on artificial intelligence. In addition, the memory 10010 can also be used to temporarily store various data that have been output or are to be output.

[0209] In some embodiments, the processor 10020 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other chips. The processor 10020 is generally used to control the overall operation of the Chinese herbal medicine efficacy evaluation device 10000, such as performing control and processing related to data interaction or communication with the Chinese herbal medicine efficacy evaluation device 10000. In this embodiment, the processor 10020 is used to run the program code stored in the memory 10010 or process data.

[0210] The network interface 10030 may include a wireless network interface or a wired network interface, and the network interface 10030 is generally used to establish a communication link between the Chinese medicinal material efficacy evaluation device 10000 and other Chinese medicinal material efficacy evaluation devices. For example, the network interface 10030 is used to connect the Chinese medicinal material efficacy evaluation device 10000 to an external terminal through a network, and to establish a data transmission channel and a communication link between the Chinese medicinal material efficacy evaluation device 10000 and the external terminal. The network can be a wireless or wired network such as an intranet, the Internet, the Global System of Mobilecommunication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G network, Bluetooth, Wi-Fi, etc.

[0211] It should be pointed out that Figure 6 Only a Chinese medicinal material efficacy evaluation device having components 10010 - 10030 is shown, but it should be understood that it is not required to implement all of the shown components, and more or fewer components may be implemented instead.

[0212] In this embodiment, the artificial intelligence-based Chinese medicinal material efficacy evaluation method stored in the memory 10010 can also be divided into one or more program modules and executed by one or more processors (such as processor 10020) to complete the embodiment of the present invention.

[0213] Embodiment 7

[0214] A computer-readable storage medium is also provided in an embodiment of the present invention, on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the method for evaluating the efficacy of traditional Chinese medicine based on artificial intelligence in the embodiment are implemented.

[0215] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the computer-readable storage medium can be an internal storage unit of the Chinese herbal medicine efficacy evaluation device, such as a hard disk or memory of the Chinese herbal medicine efficacy evaluation device. In other embodiments, the computer-readable storage medium can also be an external storage device of the Chinese herbal medicine efficacy evaluation device, such as a plug-in hard disk equipped on the Chinese herbal medicine efficacy evaluation device, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the computer-readable storage medium can also include both the internal storage unit of the Chinese herbal medicine efficacy evaluation device and its external storage device. In this embodiment, the computer-readable storage medium is generally used to store the operating system and various application software installed in the Chinese herbal medicine efficacy evaluation device, such as the program code of the Chinese herbal medicine efficacy evaluation method based on artificial intelligence in the embodiment. In addition, the computer-readable storage medium can also be used to temporarily store various types of data that have been output or are to be output.

[0216] Embodiment 8

[0217] An embodiment of the present invention further provides a computer program product, including a computer program, which implements the method in the above embodiment when executed by a processor.

[0218] Obviously, those skilled in the art should understand that the modules or steps of the above-mentioned embodiments of the present invention can be implemented by a general-purpose Chinese medicinal material efficacy evaluation device, they can be concentrated on a single Chinese medicinal material efficacy evaluation device, or distributed on a network composed of multiple Chinese medicinal material efficacy evaluation devices, optionally, they can be implemented by executable program codes of the Chinese medicinal material efficacy evaluation device, so that they can be stored in a storage device and executed by the Chinese medicinal material efficacy evaluation device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0219] It should be noted that the above are only preferred embodiments of the present invention, and the patent protection scope of the present invention is not limited thereto. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for evaluating the efficacy of traditional Chinese medicine based on artificial intelligence, characterized in that: include: S1. Data collection and annotation: Data collection and data annotation are performed at the Chinese medicinal materials production base. The collected data attributes include the growth environment of the medicinal materials and the results of component analysis. The collected data attributes are stored in a structured JSON format; S2, data expansion: The data is expanded using a multilinear interpolation algorithm based on feature coupling, and new data points are generated by evaluating the coupling between input data features; S3, feature extraction model training: extract features from the expanded data through a six-layer fully connected neural network; S4, feature dimension reduction model training: The extracted features are reduced in dimension using an autoencoder based on inverse reconstruction, and the encoding quality and decoding accuracy are optimized by minimizing the dependencies between different features; S5, classifier model training: Use the support vector machine algorithm based on quantum state mapping to classify the reduced-dimensional data; S6. Evaluation of the efficacy of Chinese medicinal materials: Use the trained model to process new Chinese medicinal materials samples to evaluate their efficacy.

2. The method for evaluating the efficacy of Chinese medicinal materials according to claim 1, characterized in that: The specific steps of data expansion include: S201, performing feature coupling analysis on the collected raw data, identifying the interdependence between the features, and obtaining coupling analysis results; S202, according to the coupling degree analysis result, selecting features with high coupling degree as selected coupling features for subsequent data generation; S203, during the data generation process, setting adaptive oscillation parameters to control the amplitude and direction of interpolation; S204, performing multiple linear interpolation using the selected coupling features and the set adaptive oscillation parameters to generate a new data set, wherein the interpolation function is dynamically adjusted according to the coupling degree of the features to generate data that conforms to the actual distribution; S205, evaluating the quality of the new data set by similarity with the statistical distribution of the original data, and adjusting the adaptive oscillation parameters according to the evaluation results to optimize the data generation process; S206 , after the adaptive oscillation parameter is adjusted, the data generation process is repeatedly iterated until a preset number of iterations is met.

3. The method for evaluating the efficacy of Chinese medicinal materials according to claim 1, characterized in that: The specific steps of the feature dimensionality reduction model training include: S301, initializing the parameters of the neural network in the feature extraction model, the parameters including the weight and bias parameters of the neural network; S302, after the parameters of the neural network are initialized or optimized and updated, when the input data passes through each layer of the neural network, a linear transformation and a nonlinear activation function are performed to output a corresponding feature representation; S303, using an exploratory gradient descent update strategy to predict gradient changes, and adjusting the step size and direction of the gradient update according to the prediction results to optimize and update the parameters of the neural network; S304: Repeat the above steps S301-S303 until the preset stop iteration condition is met, and the model training is completed.

4. The method for evaluating the efficacy of Chinese medicinal materials according to claim 1, characterized in that: The specific steps of the feature dimensionality reduction model training include: S401, initializing the parameters of the autoencoder algorithm, the parameters including the weight and bias parameters of the autoencoder; S402, in the forward propagation process, the input data is first converted into a representation of a low-dimensional space by the encoder, and then the representation of the low-dimensional space is reconstructed into output data with the same dimension as the original data by the decoder; S403, using the inverse reconstruction loss function to calculate the inverse reconstruction loss of the autoencoder; S404, using a gradient descent algorithm to update the weights and bias parameters of the encoder and decoder according to the gradient calculated by the inverse reconstruction loss function; S405: Repeat the above steps S401-S404 until the preset stop iteration condition is met, and the model training is completed.

5. The method for evaluating the efficacy of Chinese medicinal materials according to claim 1, characterized in that: The specific steps of the classifier model training include: S501, converting the data after feature dimension reduction into quantum state representation, each data point is encoded as a quantum bit sequence, and its probability amplitude is used to represent the characteristics of the data; S502, using a quantum state mapping kernel function to calculate the similarity between two data points through the inner product of the quantum state; S503, optimizing the classifier model using an optimization objective function based on density estimation; S504: Based on the above optimization objective function, the weights and bias parameters of the classifier model are optimized through quantum state mapping and quantum computing to achieve the optimal classification effect and complete the classifier model training.

6. A Chinese herbal medicine efficacy evaluation device based on artificial intelligence, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein: The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 5.

Images