Breeding selection method for cattle breeding and system thereof

Abstract

The application relates to the field of intelligent screening, and particularly discloses a breeding screening method and system for cattle breeding. The method mines the correlation feature distribution information between the semantic correlation features of each data item in the basic information of a cattle object and the implied correlation features of each data item in production data by using computer technology and bioinformatics methods, so as to screen the cattle objects suitable for breeding, thereby providing a scientific basis for cattle breeding, helping farmers to better select and breed the cattle objects, and improving breeding efficiency and cattle breeding benefits.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent screening, and more particularly, to a breeding screening method and system for cattle breeding. BACKGROUND

[0002] The breeding industry has been one of the important industries of human beings. In modern cattle breeding, the breeding and mating of cattle is a very important link in animal husbandry, which has important significance for improving the production performance and vitality of cattle. However, the traditional breeding method is mainly based on human experience and observation, which has inherent defects and limitations, such as difficulty in quickly and accurately analyzing a large amount of data, low screening efficiency, strong subjectivity, etc., which cannot fully meet the needs of modern cattle breeding.

[0003] Therefore, an optimized breeding screening scheme for cattle breeding is expected. SUMMARY

[0004] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a breeding screening method and system for cattle breeding, which mines the correlation feature distribution information between the semantic correlation features of each data item in the basic information of the cattle object and the implicit correlation features of each data item in the production data, so as to screen the cattle objects suitable for mating, thereby providing a scientific basis for the mating of cattle, helping farmers to better select and breed cattle objects, and improving the breeding efficiency and cattle breeding income.

[0005] According to one aspect of the present application, a breeding screening method for cattle breeding is provided, which comprises:

[0006] receiving basic information and production data of a to-be-screened cattle object from a cattle farm data management platform, wherein the basic information includes breed, gender, age, weight, body type and bloodline, and the production data includes milk yield, meat yield and reproduction rate;

[0007] performing feature extraction and feature correlation coding on the basic information and production data of the to-be-screened cattle object to obtain a classification feature matrix;

[0008] passing the classification feature matrix through a classifier to obtain a classification result, wherein the classification result is used to represent whether the breeding value of the to-be-screened cattle object exceeds a predetermined threshold.

[0009] According to another aspect of the present application, a breeding screening system for cattle breeding is provided, which comprises:

[0010] The information collection module is configured to receive basic information and production data of the to-be-screened cattle object from a cattle farm data management platform, wherein the basic information includes breed, gender, age, weight, body shape and bloodline, and the production data includes milk yield, meat yield and reproduction rate.

[0011] The feature extraction and correlation coding module is configured to perform feature extraction and feature correlation coding on the basic information and production data of the to-be-screened cattle object to obtain a classification feature matrix.

[0012] The classification result generation module is configured to pass the classification feature matrix through a classifier to obtain a classification result, which is used to indicate whether the breeding value of the to-be-screened cattle object exceeds a predetermined threshold.

[0013] Compared with the prior art, the breeding selection method and system for cattle breeding provided by the present application can mine the correlation feature distribution information between the semantic correlation features of each data item in the basic information of the cattle object and the implicit correlation features of each data item in the production data, so as to screen the cattle object suitable for breeding, thereby providing a scientific basis for cattle breeding, helping farmers to better select and breed cattle objects, and improving breeding efficiency and cattle breeding income. BRIEF DESCRIPTION OF DRAWINGS

[0014] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description of embodiments of the present application, taken in conjunction with the accompanying drawings. The drawings provided in the specification and the embodiments of the present application together serve to provide a further understanding that enables others skilled in the art to make or use the present application. The drawings provided are for illustrative purposes and are not intended to limit the present application. In the drawings, the same reference numerals generally refer to the same components or steps throughout the drawings.

[0015] Figure 1 A flowchart of the breeding selection method for cattle breeding according to the embodiments of the present application;

[0016] Figure 2 A system architecture diagram of the breeding selection method for cattle breeding according to the embodiments of the present application;

[0017] Figure 3 A flowchart of sub-step S120 of the breeding selection method for cattle breeding according to the embodiments of the present application;

[0018] Figure 4 A flowchart of sub-step S121 of the breeding selection method for cattle breeding according to the embodiments of the present application;

[0019] Figure 5 A block diagram of the breeding selection system for cattle breeding according to the embodiments of the present application. DETAILED DESCRIPTION

[0020] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part but not all of the embodiments of the present application, and the present application can be implemented in many different forms. Therefore, the attached drawings should not be considered as limiting the present application, and the present application should cover all changes falling into the principles and the scope of the present application.

[0021] Figure 1 A flowchart of a breeding selection method for cattle breeding according to an embodiment of the present application. Figure 2 A system architecture diagram of a breeding selection method for cattle breeding according to an embodiment of the present application. As shown in Figure 1 and Figure 2 As shown in the drawings, the breeding selection method for cattle breeding according to an embodiment of the present application comprises the steps of: S110, receiving basic information and production data of a to-be-selected cattle object from a cattle farm data management platform, wherein the basic information comprises breed, gender, age, weight, body type and bloodline, and the production data comprises milk yield, meat yield and reproduction rate; S120, performing feature extraction and feature association coding on the basic information and production data of the to-be-selected cattle object to obtain a classification feature matrix; and S130, inputting the classification feature matrix into a classifier to obtain a classification result, wherein the classification result is used to indicate whether a breeding value of the to-be-selected cattle object exceeds a predetermined threshold.

[0022] More specifically, in step S110, the basic information and production data of the to-be-selected cattle object are received from the cattle farm data management platform, wherein the basic information comprises breed, gender, age, weight, body type and bloodline, and the production data comprises milk yield, meat yield and reproduction rate. It can be understood that each data item in the basic information of the to-be-selected cattle object reflects the growth and development status and genetic gene type of the cattle object, which is of great significance for evaluating the genetic quality and suitability of the cattle; the production data of the to-be-selected cattle object reflects the production potential and production benefit of the cattle, and is also one of the key indicators for judging whether the cattle is suitable for breeding. In a specific example, first, the selection criteria and selection indicators can be determined according to the management needs of the cattle farm, for example, requiring high milk yield, good reproduction rate, moderate weight, etc. Then, using data mining and analysis techniques, the to-be-selected cattle object is selected and sorted, and the cattle that meet the criteria are selected, and the selected cattle are further evaluated and analyzed, including comprehensive evaluation of their health status, genetic background, reproductive capacity, etc., and then the selected cattle are identified and recorded for subsequent management and tracking, and finally the selection results are fed back to the customers and cattle farm management personnel so that they can understand the production situation and management effect of the cattle.

[0023] The cattle farm data management platform is a management system based on computer technology and data analysis technology, primarily used for data collection, storage, analysis, and management in cattle farms. This platform can collect various data from the farm, including basic information about the cattle, production data, health status, and feeding management. Through data mining and analysis techniques, it analyzes and processes this data, providing useful information and decision support to cattle farm managers, enabling them to better manage and operate the farm. Simultaneously, the platform also provides data sharing and exchange functions, facilitating information exchange and cooperation between different cattle farms.

[0024] More specifically, in step S120, feature extraction and feature association encoding are performed on the basic information and production data of the cattle to be screened to obtain a classification feature matrix. Accordingly, in a specific example of this application, such as... Figure 3 As shown, the basic information and production data of the cattle to be screened are used to extract features and perform feature association encoding to obtain a classification feature matrix, including: S121, performing semantic understanding on the basic information of the cattle to be screened to obtain a semantic understanding feature vector of the basic information of the cattle; S122, extracting the implicit feature vector of the production data according to the data items; and S123, performing association encoding on the semantic understanding feature vector of the basic information of the cattle and the implicit feature vector of the production data to obtain the classification feature matrix.

[0025] In step S121, semantic understanding is performed on the basic information of the cattle to be screened to obtain a semantic understanding feature vector of the cattle's basic information. It should be understood that each data item in the basic information of the cattle to be screened is composed of individual words, and these words have contextual semantic relationships. Therefore, in order to perform semantic understanding on the basic information of the cattle to be screened, and to capture and characterize the implicit features of the cattle's genetic quality and growth and development, the technical solution of this application further segments the basic information of the cattle to be screened into words to avoid subsequent word order disorder, and then encodes it using a contextual semantic encoder containing a word embedding layer. This extracts the global contextual semantic relationship feature information of each data item in the basic information of the cattle to be screened, i.e., the semantic understanding features of the basic information of the cattle to be screened, to obtain the semantic understanding feature vector of the cattle's basic information.

[0026] Accordingly, in a specific example of this application, such as Figure 4As shown, the semantic understanding of the basic information of the cow objects to be screened to obtain the semantic understanding feature vector of the basic information of the cow objects includes: S1211, performing word segmentation on the basic information of the cow objects to be screened to transform the basic information of the cow objects to be screened into a word sequence composed of multiple words; S1212, passing the word sequence through a contextual semantic encoder containing a word embedding layer to obtain the semantic understanding feature vector of the basic information of the cow objects. Specifically, S1212 includes: using the embedding layer of the contextual semantic encoder containing the embedding layer to map each word in the word sequence into a word embedding vector to obtain a sequence of word embedding vectors; using the converter of the contextual semantic encoder containing the embedding layer to perform global contextual semantic encoding on the sequence of word embedding vectors based on the converter idea to obtain multiple global contextual semantic feature vectors; and concatenating the multiple global contextual semantic feature vectors to obtain the semantic understanding feature vector of the basic information of the cow objects. The contextual semantic encoder is a technique for natural language processing that can transform input text into a vector representation for subsequent processing. The principle behind this technique is based on deep learning models such as Recurrent Neural Networks (RNNs) or their variants, Long Short-Term Memory Networks (LSTMs). It encodes each word of the input text to obtain a vector representation. During this process, the model considers contextual information—the influence of preceding and following words on the current word—to better capture the semantic information of the text. Specifically, the context semantic encoder transforms each word in the input text into a vector representation, which is then processed by models such as recurrent neural networks or LSTMs to obtain a final overall vector representation. This vector representation can be used in tasks such as text classification, sentiment analysis, and machine translation. In summary, the context semantic encoder is a technique that converts natural language into vector representations, helping computers better understand and process natural language.

[0027] The embedding layer is part of the contextual semantic encoder, used to transform the input text into a vector representation. Below is an explanation of how the embedding layer works. The main function of the embedding layer is to transform the discretized input text into a continuous vector representation, facilitating subsequent processing and analysis. Specifically, the embedding layer maps each word to a vector representation, which can be seen as the word's position in the semantic space.

[0028] Next, in S122, considering the correlation between various data items in the production data of the cattle to be screened, including milk production, meat production, and reproductive rate, in order to extract the implicit correlation features between these data to evaluate the production potential and efficiency of the cattle to be screened, and thus select superior cattle for breeding, in the technical solution of this application, the production data is arranged into an input vector according to the data items and sample dimensions, and then a feature extractor based on a multilayer perceptron model is used to obtain the production data implicit feature vector. In a specific example, the production data is arranged into an input vector according to the data items and sample dimensions; and the input vector is used to obtain the production data implicit feature vector by a feature extractor based on a multilayer perceptron model.

[0029] In one possible implementation, the production data is arranged into an input vector according to the data items and sample dimensions. This includes: determining the production data items to be collected, such as weight, yield, feed intake, etc.; determining the samples to be collected, such as the production data of a cow on a certain day; arranging the collected production data according to the data items, such as arranging the weight, yield, feed intake, etc., in sequence; arranging the arranged production data according to the sample dimensions, such as arranging the weight, yield, feed intake, etc. of a cow on a certain day in order, to form an input vector; repeating the above steps to collect and arrange the production data of all the required samples to form multiple input vectors.

[0030] Next, the feature extractor based on the multilayer perceptron model can extract high-dimensional implicit correlation features between various data items in the production data, thereby more fully representing the production capacity of the cattle to be screened. Specifically, the multilayer perceptron (MLP) is a common feedforward neural network composed of multiple neuron layers, each fully connected. MLPs can be used for tasks such as classification and regression, and are one of the most fundamental models in deep learning. The principle of MLP is based on the working principle of neurons. Each neuron receives the input signal, performs a weighted summation of the input signal, adds a bias term, and then performs a nonlinear transformation through an activation function, finally outputting a result. MLP combines multiple neuron layers together, continuously performing nonlinear transformations on the input signal to obtain a more complex model. Specifically, the input layer of an MLP receives input data and then passes the input data to the first hidden layer. Each hidden layer consists of multiple neurons, each of which performs a weighted summation of the input signal, adds a bias term, and then performs a nonlinear transformation through an activation function. The connections between hidden layers are also fully connected; the output of each hidden layer is passed to the next hidden layer until the output of the last hidden layer is passed to the output layer. The output layer typically consists of one or more neurons, and the output of each neuron is the final prediction. The training process of an MLP usually uses the backpropagation algorithm, which calculates the error between the predicted and true values ​​and then propagates the error back to each neuron using a chain rule, thereby updating the weights and biases between neurons and ultimately making the model's predictions more accurate. In summary, a multilayer perceptron (MLP) is a neuron-based feedforward neural network model that uses nonlinear transformations between multiple neuron layers to obtain a more complex model that can be used for tasks such as classification and regression.

[0031] Then, in step S123, after obtaining the semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data, the two are further correlated and encoded to represent the correlation feature distribution information between the semantic correlation features of each data item in the basic information of the cattle object to be screened and the high-dimensional implicit correlation features of each data item in the production data. That is, the correlation feature information between the growth state and genetic type features of the cattle object to be screened and the implicit features of the production state, so as to obtain the classification feature matrix. The associative encoding is an unsupervised learning method. Its principle is to map the input data onto codewords in a low-dimensional space, so that similar input data are closer in the codeword space, while dissimilar input data are farther apart. The main idea of ​​associative encoding is to enable codewords to effectively represent the structural information of the input data through an adaptive learning process, thereby achieving data compression and dimensionality reduction. The implementation process of associative encoding can be divided into two stages: a learning stage and an encoding stage. In the learning stage, a linear transformation matrix is ​​learned by minimizing the reconstruction error, mapping the input data to the codeword space. During the encoding phase, the corresponding codeword representation is obtained by multiplying the input data by the learned linear transformation matrix. The distance between codewords can be measured using metrics such as Euclidean distance or cosine similarity. This association coding can be applied to multiple fields, such as image processing, natural language processing, and recommender systems. In image processing, association coding can be used for image compression and dimensionality reduction; in natural language processing, it can be used for word vector learning and text classification; and in recommender systems, it can be used for user and item representation and recommendation.

[0032] In one possible implementation, firstly, the basic information and feature vectors of the cattle to be screened, based on production data, are determined. Machine learning or deep learning models can be used for semantic understanding and feature extraction. Next, the obtained feature vectors are associatively encoded, using methods such as hash encoding or one-hot encoding. Then, the encoded feature vectors are combined into a classification feature matrix, which can be manipulated using libraries such as NumPy or Pandas. Finally, the resulting classification feature matrix is ​​used for cattle screening and classification.

[0033] In another possible implementation, firstly, the basic information and production data of the cattle are preprocessed, including data cleaning, feature extraction, and feature engineering. Then, the preprocessed data is modeled using machine learning or deep learning models for classification and prediction. Next, the classification results are correlated with the basic information of the cattle to be screened, and data manipulation can be performed using libraries such as pandas. Finally, the classification results for the cattle to be screened are obtained.

[0034] In a specific example of this application, the semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data are correlated and encoded using the following formula to obtain a classification feature matrix; wherein, the formula is: Where V m This represents the semantic understanding feature vector of the basic information of the cow object. V represents the transpose of the semantic understanding feature vector of the basic information of the cow object. n M represents the implicit feature vector of the production data, and M represents the classification feature matrix. This represents vector multiplication.

[0035] More specifically, in step S130, the classification feature matrix is ​​processed by a classifier to obtain a classification result, which indicates whether the breeding value of the cattle to be screened exceeds a predetermined threshold. The classification feature matrix is ​​further processed by a classifier to obtain a classification result indicating whether the breeding value of the cattle to be screened exceeds the predetermined threshold. In the technical solution of this application, the semantic understanding feature vector of the basic information of the cattle object expresses the textual context-related semantics of the basic information of the cattle to be screened, while the implicit feature vector of the production data expresses the implicit features of the production data. Due to the correlation between the basic information of the cattle to be screened and the production data itself, the semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data also have feature correlation. Therefore, when associating and encoding the semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data, the semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data are associated position by position to obtain the feature value of each position of the classification feature matrix. Although the process and long-range association information of the semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data can be obtained, it also leads to insufficient dimensionality discrimination of the overall feature distribution composed of the feature values ​​of each position of the classification feature matrix in the high-dimensional feature space under the probability density representation. This affects the class probability expression of the classification feature matrix and the accuracy of the classification result obtained by the classifier. Therefore, the applicant of this application performs Gaussian probability density manifold surface dimensionality orthogonalization on the classification feature matrix, for example, denoted as M, as follows:

[0036]

[0037] Where m i,j Let m be the eigenvalue at position (i,j) of the classification feature matrix, where μ and σ are the mean and standard deviation of the set of eigenvalues ​​at each position in the classification feature matrix, respectively. i,j' is the eigenvalue at position (i,j) of the optimized classification feature matrix. Here, the unit tangent vector magnitude and unit normal vector magnitude of the surface are characterized by the square root of the mean and standard deviation of the high-dimensional feature set representing the manifold surface. The manifold surface of the high-dimensional feature matrix M can be orthogonally projected onto the tangent and normal planes based on the unit magnitude. This allows for dimensional reshaping of the probability density of the high-dimensional features based on the basic structure of the Gaussian feature manifold geometry. By enhancing the orthogonality of the probability density, the accuracy of the class probability expression of the optimized classification feature matrix is ​​improved, thereby improving the accuracy of the classification results obtained by the classifier. In other words, by passing the optimized classification feature matrix through a classifier to obtain the classification results, suitable cattle for breeding can be accurately selected, providing a scientific basis for cattle breeding and helping farmers to better select cattle, improve breeding efficiency, and increase cattle farming profits. Specifically, the classifier includes multiple fully connected layers and a Softmax layer cascaded with the last fully connected layer of the multiple fully connected layers. In the classification process of the classifier, the optimized classification feature matrix is ​​first projected into a vector. For example, in a specific example, the optimized classification feature matrix is ​​expanded along row or column vectors to form a classification feature vector. Then, the classification feature vector is fully encoded multiple times using multiple fully connected layers of the classifier to obtain an encoded classification feature vector. Furthermore, the encoded classification feature vector is input into the softmax layer of the classifier, that is, the softmax classification function is used to classify the encoded classification feature vector to obtain a classification label. In the technical solution of this application, the labels of the classifier include whether the breeding value of the cattle to be screened exceeds a predetermined threshold (first label) and whether the breeding value of the cattle to be screened does not exceed a predetermined threshold (second label). The classifier determines which classification label the classification feature matrix belongs to using a softmax function. It is worth noting that the first label p1 and the second label p2 here do not contain artificially defined concepts. In fact, during the training process, the computer model does not have the concept of "whether the breeding value of the cattle to be screened exceeds a predetermined threshold." It simply has two classification labels and outputs the probability of the feature under these two classification labels, i.e., the sum of p1 and p2 is one. Therefore, the classification result of whether the breeding value of the cattle to be screened exceeds the predetermined threshold is actually transformed into a binary probability distribution that conforms to natural laws through classification labels. In essence, it uses the physical meaning of the natural probability distribution of the labels, rather than the linguistic textual meaning of "whether the breeding value of the cattle to be screened exceeds the predetermined threshold".It should be understood that in the technical solution of this application, the classification label of the classifier is a detection and evaluation label for whether the breeding value of the cattle to be screened exceeds a predetermined threshold. Therefore, after obtaining the classification result, suitable cattle for breeding can be screened based on the classification result, thereby providing a scientific basis for cattle breeding. The classifier is a machine learning model used to classify data into different categories. The principle of the classifier is to learn a classification function or decision rule based on the sample features and category labels in the training dataset, used to classify new data. The training process of the classifier involves learning and optimizing the training dataset to obtain the optimal classification function or decision rule to predict the category of new data with the greatest accuracy. In summary, a classifier is a model learned based on training data, which can be used to classify new data into known categories.

[0038] The Softmax function is a commonly used activation function, primarily used in multi-class classification problems. It maps a set of arbitrary real values ​​to a probability distribution such that the probability of each real value lies within the interval [0,1], and the sum of all probabilities is 1. During training, the Softmax function is often used in conjunction with the cross-entropy loss function to calculate the difference between the predicted result and the true label. During prediction, the Softmax function transforms the model's output into a class probability distribution for classification decisions. Essentially, the Softmax function normalizes the input vector using an exponential function, ensuring that each element's value is between [0,1] and the sum of all elements is 1. Thus, we can view the input vector as a score for each class; the Softmax function transforms these scores into a probability distribution, facilitating classification tasks.

[0039] In summary, the breeding screening method for cattle breeding according to the embodiments of this application is explained. It uses computer technology and bioinformatics methods to mine the distribution information of the correlation features between the semantic association features of each data item in the basic information of cattle and the implicit association features of each data item in the production data. This information is used to screen suitable cattle for breeding, thereby providing a scientific basis for cattle breeding and helping farmers to better select and breed cattle, improve breeding efficiency and cattle breeding income.

[0040] Furthermore, a breeding screening system for cattle farming is also provided.

[0041] Figure 5 This is a block diagram of a breeding screening system for cattle farming according to an embodiment of this application. Figure 5As shown, the breeding screening system 300 for cattle breeding according to an embodiment of this application includes: an information acquisition module 310, used to receive basic information and production data of cattle to be screened from a cattle farm data management platform, wherein the basic information includes breed, sex, age, weight, body type and pedigree, and the production data includes milk production, meat production and reproductive rate; a feature extraction and association encoding module 320, used to perform feature extraction and feature association encoding on the basic information and production data of the cattle to be screened to obtain a classification feature matrix; and a classification result generation module 330, used to pass the classification feature matrix through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the breeding value of the cattle to be screened exceeds a predetermined threshold.

[0042] As described above, the breeding screening system for cattle farming according to the embodiments of this application can be implemented in various terminal devices. In one example, the breeding screening system 300 for cattle farming according to the embodiments of this application can be integrated into the terminal device as a software module and / or a hardware module. For example, the breeding screening system 300 for cattle farming can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the breeding screening system 300 for cattle farming can also be one of many hardware modules of the terminal device.

[0043] Alternatively, in another example, the breeding screening system 300 for cattle breeding and the terminal device can also be separate devices, and the breeding screening system 300 for cattle breeding can be connected to the terminal device via wired and / or wireless networks, and transmit interactive information in accordance with an agreed data format.

[0044] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for selecting breeding stock for cattle farming, characterized in that, include: The system receives basic information and production data of the cattle to be screened from the cattle farm data management platform. The basic information includes breed, sex, age, weight, body type and pedigree, and the production data includes milk production, meat production and reproductive rate. The basic information of the cattle objects to be screened is semantically understood to obtain the semantic understanding feature vector of the basic information of the cattle objects; The production data is arranged into an input vector according to the sample dimension of the data items, and the input vector is passed through a feature extractor based on a multilayer perceptron model to obtain the amount of latent feature vectors of the production data. The semantic understanding feature vector of the basic information of the cattle object and the implicit feature vector of the production data are correlated and encoded using the following formula to obtain a classification feature matrix; The formula is as follows: in This represents the semantic understanding feature vector of the basic information of the cow object. This represents the transpose of the semantic understanding feature vector of the basic information of the cow object. This represents the implicit feature vector of the production data. This represents the classification feature matrix. This represents vector multiplication; The classification feature matrix is ​​passed through a classifier to obtain a classification result, which is used to indicate whether the breeding value of the cattle to be screened exceeds a predetermined threshold. The classification feature matrix is ​​processed by a classifier to obtain a classification result, which indicates whether the breeding value of the cattle to be screened exceeds a predetermined threshold, including: The classification feature matrix is ​​optimized by performing feature distribution optimization to obtain an optimized classification feature matrix; and The optimized classification feature matrix is ​​passed through the classifier to obtain the classification result; The process of optimizing the feature distribution of the classification feature matrix to obtain an optimized classification feature matrix includes: The classification feature matrix is ​​orthogonalized to the Gaussian probability density manifold surface dimension using the following optimization formula to obtain the optimized classification feature matrix; The optimization formula is as follows: in It is the first of the classification feature matrices Location feature value and These are the mean and standard deviation of the feature value set at each position in the classification feature matrix, respectively. It is the first of the optimized classification feature matrices. The characteristic value of the location.

2. The method for breeding and screening of cattle according to claim 1, characterized in that, The basic information of the cattle objects to be screened is semantically understood to obtain a semantic understanding feature vector of the basic information of the cattle objects, including: The basic information of the cattle objects to be screened is segmented into words to transform it into a word sequence composed of multiple words; and The word sequence is passed through a contextual semantic encoder containing a word embedding layer to obtain a semantic understanding feature vector of basic information about the cow object.

3. The breeding screening method for cattle breeding according to claim 2, characterized in that, The word sequence is passed through a contextual semantic encoder containing a word embedding layer to obtain a semantic understanding feature vector of basic information about the cow object, including: The word embedding layer of the context semantic encoder containing the word embedding layer is used to map each word in the word sequence into a word embedding vector to obtain a sequence of word embedding vectors; The sequence of word embedding vectors is subjected to global contextual semantic encoding based on the converter concept using the contextual semantic encoder containing the word embedding layer to obtain multiple global contextual semantic feature vectors; and The multiple global context semantic feature vectors are concatenated to obtain the basic information semantic understanding feature vector of the cow object.

4. The breeding screening method for cattle breeding according to claim 3, characterized in that, The optimized classification feature matrix is ​​passed through a classifier to obtain classification results. These classification results indicate whether the breeding value of the cattle to be screened exceeds a predetermined threshold, including: The optimized classification feature matrix is ​​expanded into a classification feature vector based on row vectors or column vectors; The classification feature vector is fully encoded using multiple fully connected layers of the classifier to obtain an encoded classification feature vector; and The encoded classification feature vector is passed through the Softmax classification function of the classifier to obtain the classification result.

5. A breeding screening system for cattle farming, characterized in that, include: The information collection module is used to receive basic information and production data of the cattle to be screened from the cattle farm data management platform. The basic information includes breed, sex, age, weight, body type and pedigree, and the production data includes milk production, meat production and reproductive rate. The feature extraction and association encoding module is used to extract and associate features with the basic information and production data of the cattle objects to be screened to obtain a classification feature matrix. This includes: performing semantic understanding on the basic information of the cattle objects to be screened to obtain a semantic understanding feature vector of the basic information of the cattle objects; arranging the production data into an input vector according to data items and sample dimensions, and passing the input vector through a feature extractor based on a multilayer perceptron model to obtain a latent feature vector of the production data; and associating and encoding the semantic understanding feature vector of the basic information of the cattle objects and the latent feature vector of the production data with the following formula to obtain a classification feature matrix. The formula is as follows: in This represents the semantic understanding feature vector of the basic information of the cow object. This represents the transpose of the semantic understanding feature vector of the basic information of the cow object. This represents the implicit feature vector of the production data. This represents the classification feature matrix. This represents vector multiplication; The classification result generation module is used to pass the classification feature matrix through a classifier to obtain a classification result, which indicates whether the breeding value of the cattle to be screened exceeds a predetermined threshold. The classification feature matrix is ​​processed by a classifier to obtain a classification result, which indicates whether the breeding value of the cattle to be screened exceeds a predetermined threshold, including: The classification feature matrix is ​​optimized by performing feature distribution optimization to obtain an optimized classification feature matrix; and The optimized classification feature matrix is ​​passed through the classifier to obtain the classification result; The process of optimizing the feature distribution of the classification feature matrix to obtain an optimized classification feature matrix includes: The classification feature matrix is ​​orthogonalized to the Gaussian probability density manifold surface dimension using the following optimization formula to obtain the optimized classification feature matrix; The optimization formula is as follows: in It is the first of the classification feature matrices Location feature value and These are the mean and standard deviation of the feature value set at each position in the classification feature matrix, respectively. It is the first of the optimized classification feature matrices. The characteristic value of the location.

Images