CN115809601A - Sedimentary rock structure background distinguishing method - Google Patents
Sedimentary rock structure background distinguishing method Download PDFInfo
- Publication number
- CN115809601A CN115809601A CN202211557450.1A CN202211557450A CN115809601A CN 115809601 A CN115809601 A CN 115809601A CN 202211557450 A CN202211557450 A CN 202211557450A CN 115809601 A CN115809601 A CN 115809601A
- Authority
- CN
- China
- Prior art keywords
- neural network
- deep neural
- network model
- function
- data set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the technical field of artificial intelligence and geology, and particularly discloses a sedimentary rock structure background distinguishing method which comprises the core steps of constructing a data set, constructing a deep neural network model, training the deep neural network model, distinguishing a rock structure background and the like. The method is mainly realized by a designed residual error deep neural network model based on a self-attention mechanism; the provided deep neural network model is built and trained according to the content of the corresponding main trace elements, and the related parameters can be adjusted and optimized according to the related method provided by the invention to obtain the optimal models corresponding to different tasks; on the basis of completing model training and debugging, a user can directly drive the deep neural network model to read the main trace element content data of the target sample, and further automatically obtain the construction background information of the target sample through the model, assist the user in completing relevant research and improve the working efficiency.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence and geology, and particularly relates to a sedimentary rock structure background judging method.
Background
The exploration of the relationship between the geochemical characteristics of rocks and the tectonic background is an important research direction in the field of geochemistry, and the principle is that the geochemical elements of rocks have certain indication effect on the tectonic background. The research on the structural background of the rock is helpful for finding out the information of geological structure, evolution process, ore bed distribution and the like in a specific area, and has great strategic significance on the utilization of national natural resources and the environmental protection. The mainstream distinguishing method at present is to utilize the geochemical analysis results of the whole rock, including the data of the composition of major elements, trace elements and isotopes, and combine the distinguishing diagram provided by the previous research to analyze so as to distinguish the structural background of the rock. The method has the characteristics of solid theoretical basis, rich research results, simple expression mode and easy use, but along with the deepening of geochemical research, the method has strong experience and subjectivity, various distinguishing schemes, limitation in the application range of each diagram, contradiction between different scheme results and the like.
For the above technical problems, at present, domestic efforts are mainly made to make more discriminant diagrams in order to obtain better analysis accuracy, and some studies are also made by using simple machine learning such as: algorithms such as a decision tree, a support vector machine and the like are used for learning and classifying geochemical characteristics, the number of the used model layers is small, the algorithm structure is simple, and the judgment accuracy is poor. Foreign countries often adopt a numerical analysis method to model the chemical characteristic data of the rock so as to optimize the effect of the discriminant diagram, and the method aims at single specific rocks such as: sedimentary rock, metamorphic rock and the like have obvious optimization effects, and the structural background of single rock can be rapidly distinguished; however, in engineering practice, different rocks need to be modeled and parameters need to be adjusted respectively, and geochemical characteristic data needs to be cleaned and manually screened, so that the method still has the problems of low robustness, poor flexibility, complex process and extreme dependence on manpower.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a sedimentary rock structure background judging method.
The invention provides a sedimentary rock structure background judging method, which comprises the following steps:
the method comprises the steps of obtaining original sample data of a rock, classifying the original sample data according to a rock structure background, simultaneously extracting geochemical characteristics of the rock in the original sample data to construct an original data set, and labeling a structure background label of the sample according to a classification result;
taking the original data set as a training set and a testing set;
inputting the original data set into a deep neural network model to obtain a data set containing semantic information;
calculating the sample category score of the data set containing the semantic information to obtain the prediction probability of the data set containing the semantic information corresponding to different rock structure backgrounds, namely completing the construction of a deep neural network model;
and 4, judging the rock structure background by using the trained deep neural network model.
Further, the deep neural network model construction includes:
performing dimensionality-increasing processing on the original data set through a full connection layer and an activation function to obtain a dimensionality-increasing data set;
performing nonlinear calculation on the dimensionality-increased data set by utilizing two self-attention residual error modules to obtain a feature vector with higher semantic information;
reducing the dimension of the characteristic vector through a full connection layer, an activation function and a self-attention residual error module to densify the contained characteristics;
calculating by utilizing a full connection layer to obtain a sample category score, and converting the sample category score into the prediction probability of the target construction background through a Softmax function;
adjusting model parameters by using a training set, evaluating training errors by using a cross soil moisture loss function as a loss function, performing iterative optimization on the model parameters by using a random gradient descent optimizer, and verifying training precision on a test set to obtain an optimal model;
and reading the rock structure background data by using the deep neural network model, and automatically outputting the most possible structure background category of the target sample by using the model to realize discrimination operation.
The further scheme is that the normalization formula of the content of the main trace elements is as follows:
in the formula, X i Mainly trace elements, mu i Sample mean of major trace elements, S i Is the standard deviation of the sample, X scale Is a normalized result.
Further, the self-attention residual module is configured to: calculating feature weights of different feature positions from the feature vectors of the self-attention residual error module through a full connection layer and a Softmax function, wherein the sum of the weights is 1; multiplying the feature weight by the feature vector element by element to obtain a weighted feature; transmitting the weighted features into a full connection layer and a Relu activation function of which the input and output channels are both 100, and further performing nonlinear calculation; adding the obtained characteristic vector element by element with the input characteristic vector of the self-attention residual error module through residual error connection to obtain an output characteristic vector with the length of 100;
in a further aspect, the number of input and output channels of the self-attention residual error module is the same.
Further, the nonlinear calculation is calculated by using a Relu function, and the expression is as follows:
f(x)＝max(0,x)
relu function as the activation function of a neuron defines the linear transformation w of the neuron T The result of the non-linear output of x + b, i.e., for an input vector x from the neural network of the previous layer from a neuron, the neuron using the Relu function will output max (0, w) T x+b)
And send the result to the next layer of neurons or as output for the entire neural network.
In a further aspect, the higher semantic information inputs a probability value for each constructed environment for a set of samples.
The further scheme is that the Softmax function is a normalized exponential function, and the function expression of the Softmax function is as follows:
in the formula, z i The output value of the ith node is shown, C is the number of the output nodes, e is a natural base number, and the classification scores of a plurality of classification outputs can be converted into a range of [0,1 ] by a Softmax function]And a probability distribution of 1.
Compared with the prior art, the invention has the beneficial effects that:
the method utilizes the nonlinear fitting capability of the deep neural network to adaptively learn the latent relation between different main trace element contents and the construction background category through sample data, and the precision and the accuracy of the method are remarkably improved compared with the traditional artificial graphical discrimination method, mathematical analysis and modeling method.
The self-attention residual error module suitable for rock structure background judgment is used, the weight can be obtained by utilizing the self-calculation of the features for weighted calculation, the extraction efficiency of the features is improved, and the judgment performance and robustness are also improved; meanwhile, due to the characteristics of the self-attention residual error module, the features of different depths can be added element by element, the defects of gradient disappearance and gradient explosion caused by overlarge depth of a traditional deep neural network model are overcome, and the convergence efficiency during training is greatly improved.
The rock structure background discrimination method provided by the invention has stronger flexibility. When the deep neural network model has enough rock geochemical characteristic data for training, the source regions of more different types of rocks can be distinguished, the traditional distinguishing mode of 'single type single operation' is broken through, and the distinguishing flexibility and the application range are greatly improved.
The automatic construction background judgment is carried out by utilizing the deep neural network model, so that the labor intensity of a user can be greatly reduced, and the manpower is liberated; the judgment can be realized only by collecting partial geochemical content data of the rock, redundant operation is reduced, the operation efficiency is higher, and the speed is higher.
Drawings
The invention is illustrated and described only by way of example and not by way of limitation in the scope of the invention as set forth in the following drawings, in which:
FIG. 1 is a schematic diagram of a self-attention mechanism-based residual deep neural network according to the present invention;
FIG. 2 is a graph illustrating a learning rate decay curve according to the present invention;
FIG. 3 is a graph illustrating a loss function curve according to the present invention;
FIG. 4 is a schematic diagram of a training set discrimination accuracy curve according to the present invention;
FIG. 5 is a schematic diagram of a test set discrimination accuracy curve according to the present invention;
FIG. 6 is a schematic diagram of a raw data set constructed in accordance with the present invention;
FIG. 7 is a schematic diagram of a partitioned training set according to the present invention;
FIG. 8 is a schematic diagram of a partitioned test set according to the present invention;
fig. 1 is a diagram of a residual error deep neural network structure based on an attention mechanism for discriminating a rock structure background, which is provided by the present invention, and inputs geochemical characteristics of a sample, and outputs a structural background discrimination result of a target sample through a full connection layer, an activation function, data upscaling characteristic extraction, data downscaling characteristic densification and prediction discrimination;
fig. 2 is a graph of a learning rate decay curve of the residual deep neural network provided by the present invention, that is, a learning rate parameter adjustment process of the network model in a training process;
FIG. 3 is a graph of a loss function of the residual deep neural network provided by the present invention, showing the error variation trend of the network model in the training process;
FIG. 4 is a graph of the discrimination accuracy of the training set of the residual deep neural network provided by the present invention, which shows the accuracy variation trend of the network model on the training set;
fig. 5 is a test set discrimination accuracy curve diagram of the residual deep neural network provided by the present invention, which shows an accuracy variation trend of the network model on the test set.
Detailed Description
In order to make the objects, technical solutions, design methods, and advantages of the present invention more apparent, the present invention will be further described in detail by specific embodiments with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
With the development of big data and artificial intelligence, various mature models have been widely applied to various basic research fields and have good effects. The deep neural network model transforms the feature representation of the sample in the original space to a new feature space by performing layer-by-layer feature transformation on the original signal, and obtains the hierarchical feature representation by automatic learning, so that the process of extracting the sample features by means of manual experience is skipped, and the data screening and classifying efficiency is greatly improved. It is expected that deep neural networks will also yield fruitful results in the field of geology.
Screening and marking the collected open source rock geochemical characteristic data, training by using the deep neural network model of the invention, and dynamically adjusting model parameters; and perfecting the deep neural network model for users to use according to the optimal parameters. When the user judges the structural background of the rock, the original geochemical characteristic data of the rock can be input, and the model automatically gives the most possible structural background of the rock, so that the user is assisted to complete related research, and manpower and material resources are greatly saved.
In order to achieve the purpose, the invention provides a sedimentary rock structure background judging method, which comprises the following steps:
1. a background data set collection is constructed. Performing geochemical analysis on the rock to obtain original sample data or directly obtaining the original sample data from an open source library; and classifying the original sample data according to an expected construction background, extracting the content of the main trace elements to construct an original data set, dividing the original data set into a training set and a testing set according to a certain proportion, and labeling the construction background label of each group of samples. Respectively used for the training and the effect test of the model; and normalizing the contents of the main trace elements in all sample data to increase the training efficiency.
Wherein, the normalization formula of the content of the main trace elements is as follows:
in the formula, X i Mainly trace elements, mu i Sample mean of major trace elements, S i Is the standard deviation of the sample, X scale Is a normalized result.
2. And (5) constructing a model of the deep neural network. Firstly, forming a multidimensional vector by normalized main trace element content data, inputting the multidimensional vector into a deep neural network model, and increasing the dimensions of the multidimensional vector through a full-connection layer and an activation function; secondly, carrying out nonlinear calculation on the multidimensional vector by utilizing two self-attention residual error modules to obtain a feature vector with higher semantic information; reducing the dimension of the feature vector through a full connection layer, an activation function and a self-attention residual error module again to densify the contained features; and finally, calculating by utilizing a full connection layer to obtain a sample category score, and converting the sample category score into the prediction probability of the target construction background through a Softmax function. Wherein, the higher semantic information inputs the probability value of each construction environment for a group of samples; the self-attention residual module is configured to: calculating feature weights of different feature positions from the feature vectors of the self-attention residual error module through a full connection layer and a Softmax function, wherein the sum of the weights is 1; multiplying the feature weight by the feature vector element by element to obtain a weighted feature; the weighted features are transmitted into a full connection layer and a Relu activation function of which the input and output channels are both 100, and nonlinear calculation is further carried out; and adding the obtained characteristic vector and the input characteristic vector of the self-attention residual error module element by element through residual error connection to obtain an output characteristic vector with the length of 100. The nonlinear calculation is calculated by using a Relu function, and the expression is as follows:
f(x)＝max(0,x)
relu function as the activation function of a neuron defines the linear transformation w of the neuron T The result of the non-linear output of x + b, i.e., for the input vector x from the neural network of the previous layer from the neuron, the neuron using the Relu function will output max (0, w) T x+b)
And send the result to the next layer of neurons or as the output of the entire neural network;
the Softmax function is a normalized exponential function, and the function expression of the Softmax function is as follows:
in the formula, z i The output value of the ith node is shown, C is the number of the output nodes, e is a natural base number, and the classification scores of a plurality of classification outputs can be converted into a range of [0,1 ] by a Softmax function]And a probability distribution of 1.
3. And (5) model training of the deep neural network. And adjusting and optimizing the model parameters by using a training set, evaluating training errors by using a cross soil moisture loss function as a loss function, performing iterative optimization on the model parameters by using a random gradient descent optimizer, and verifying the training precision on a test set to obtain an optimal model.
4. And (5) judging the rock structure background. The user reads the rock geochemical characteristic data of the target sample by using the deep neural network model provided by the invention, and the model automatically outputs the most possible construction background category of the target sample to realize the discrimination operation.
For clear and intuitive explanation of the implementation method of the present invention, the following description will be made by taking the structural background judgment of the clastic sedimentary rock as an example. The constructional background of clastic sedimentary rocks can be mainly divided into three categories: the geochemistry characteristics of the island arc, the mainland valley and the collision zone mainly comprise the following 10 major elements: siO 2 2 、TiO 2 、Al 2 O 3 、Fe 2 O 3 、MnO、MgO、CaO、Na 2 O、K 2 O、P 2 O 5 。
The specific process is as follows:
1. constructing a data set, as shown in fig. 6 to 8, extracting 2099 sedimentary rock sample data from the international open source repository GEOROC data set, and labeling the original samples as: arc (island Arc), rift (valley splitting in continent) and Col (collision zone), extracting the content of the 10 major elements to form an original data set, and then dividing the original data set into a training set and a testing set according to the proportion of 7; and calculating the mean value and the variance of the contents of the 10 types of principal component elements in all the sample data, and normalizing the data by using the obtained result to reduce the training difficulty.
2. And constructing a deep neural network model. As shown in fig. 1, a 10-dimensional feature vector composed of 10 types of principal component element content is input as a deep neural network model, and the 10-dimensional feature vector is raised to 100 dimensions through a full connection layer and an activation function; performing nonlinear calculation on the feature vectors with raised dimensions by using two self-attention residual error modules to obtain feature vectors containing semantic information; then, reducing the feature vector containing semantic information to 50 dimensions by using a full connection layer, an activation function and a self-attention residual error module, and densifying the contained features; and finally, calculating by means of a full connection layer to obtain a sample category score, and converting the score into the prediction probability of three types of construction backgrounds including Arc (island Arc), rift (mainland valley crack) and collision area (Col) by a Softmax function to complete the construction of the deep neural network model.
3. And training the deep neural network model. As shown in fig. 1, after obtaining the deep neural network model, training the model by using the divided training set; as shown in fig. 2, cosine attenuation is adopted to reduce the learning rate with the increase of the iteration number, the initial learning rate is 0.02, and the learning momentum is 0.9; as shown in FIG. 3, the error of training is evaluated by using the cross soil moisture loss function, and it can be seen that the error decreases as the number of training times increases; iteratively optimizing the related parameters by using a random gradient descent optimizer so as to obtain the optimal parameters of the deep neural network model; as shown in fig. 4 and 5, it can be seen that the training precision increases with the increase of the training times and tends to be stable, the discrimination precision of the construction background on the training set is 97%, and the discrimination precision of the source region on the test set is 84%, and finally the optimal deep neural network model reaching the practical value level is obtained.
4. And judging the rock structure background. The optimal deep neural network model can be used by a user after being obtained, the user obtains 10 types of principal component element content data of the clastic sedimentary rock sample through various technical means, the deep neural network model is driven to read the data, a structural background judgment result of the clastic sedimentary rock sample can be obtained, and automatic judgment is achieved.
For convenience of explaining the core method and technical means of the invention, the description and the schematic drawings mainly take sedimentary rocks as an example for description, but do not represent that the patent of the invention is only suitable for judging the structural background of the sedimentary rocks, and do not influence the reading and understanding of the implementation mode of the scheme provided by the embodiment by the technical staff in the field. Based on the well-known artificial intelligence and geology, the person skilled in the art combines the description and the drawings of the description, and easily leads the patent of the invention to be applied to the application of judging the backgrounds of various rock structures under the guidance of the invention.
While embodiments of the present invention have been described above, the above description is illustrative, not exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen in order to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
Claims (8)
1. A sedimentary rock structure background discrimination method is characterized by comprising the following steps:
step 1, constructing a data set:
the method comprises the steps of obtaining original sample data of a rock, classifying the original sample data according to a rock structure background, simultaneously extracting geochemical characteristics of the rock in the original sample data to construct an original data set, and labeling a structure background label of the sample according to a classification result;
taking the original data set as a training set and a testing set;
step 2, constructing a deep neural network model:
inputting the original data set into a deep neural network model to obtain a data set containing semantic information;
calculating the sample category score of the data set containing the semantic information to obtain the prediction probability of the data set containing the semantic information corresponding to different rock structure backgrounds, namely completing the construction of a deep neural network model;
step 3, training the deep neural network model constructed in the step 2 by adopting the training set;
and 4, judging the rock structure background by using the trained deep neural network model.
2. The sedimentary rock formation background discrimination method according to claim 1, wherein the deep neural network model construction includes:
performing dimensionality increasing processing on the original data set through a full connection layer and an activation function to obtain a dimensionality increasing data set;
performing nonlinear calculation on the dimensionality-increased data set by utilizing two self-attention residual error modules to obtain a feature vector with higher semantic information;
reducing the dimension of the feature vector through a full connection layer, an activation function and a self-attention residual error module to densify the contained features;
calculating by utilizing a full connection layer to obtain a sample category score, and converting the sample category score into the prediction probability of the target construction background through a Softmax function;
adjusting model parameters by using a training set, evaluating training errors by using a cross soil moisture loss function as a loss function, performing iterative optimization on the model parameters by using a random gradient descent optimizer, and verifying training precision on a test set to obtain an optimal model;
and reading the rock structure background data by using the deep neural network model, and automatically outputting the most possible structure background category of the target sample by using the model to realize discrimination operation.
3. The sedimentary rock formation background discrimination method according to claim 2, wherein the normalization formula of the content of the primary trace elements is as follows:
in the formula, X i Mainly trace elements, mu i Sample mean of major trace elements, S i Is the standard deviation of the sample, X scale Is a normalized result.
4. A sedimentary rock formation background discrimination method according to claim 2, wherein the self-attention residual module is configured to: calculating feature weights of different feature positions from the feature vectors of the self-attention residual error module through a full connection layer and a Softmax function, wherein the sum of the weights is 1; multiplying the feature weight by the feature vector element by element to obtain a weighted feature; transmitting the weighted features into a full connection layer and a Relu activation function of which the input and output channels are both 100 again, and performing nonlinear calculation; and adding the obtained feature vector and the input feature vector of the self-attention residual error module element by element through residual error connection to obtain an output feature vector with the length of 100.
5. The sedimentary rock formation background discrimination method according to claim 4, wherein the Softmax function is a normalized exponential function, and the function expression of the Softmax function is as follows:
in the formula, z i The output value of the ith node is shown, C is the number of the output nodes, e is a natural base number, and the classification scores of a plurality of classification outputs can be converted into a range of [0,1 ] by a Softmax function]And a probability distribution of 1.
6. The method as claimed in claim 4, wherein the number of input/output channels of the self-attention residual error module is the same.
7. A sedimentary rock formation background discrimination method according to claim 4, characterized in that the nonlinear calculation is calculated by using a Relu function, and the expression thereof is as follows:
f(x)＝max(0,x)
relu function as the activation function of a neuron defines the linear transformation w of the neuron T The result of the non-linear output of x + b, i.e., for the input vector x from the neural network of the previous layer from the neuron, the neuron using the Relu function will output max (0, w) T x+b)
And send the result to the next layer of neurons or as output for the entire neural network.
8. A sedimentary rock formation background discrimination method according to claim 2, wherein the higher semantic information is a probability value for each formation environment inputted for a group of samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211557450.1A CN115809601A (en) | 2022-12-06 | 2022-12-06 | Sedimentary rock structure background distinguishing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211557450.1A CN115809601A (en) | 2022-12-06 | 2022-12-06 | Sedimentary rock structure background distinguishing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115809601A true CN115809601A (en) | 2023-03-17 |
Family
ID=85485045
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211557450.1A Pending CN115809601A (en) | 2022-12-06 | 2022-12-06 | Sedimentary rock structure background distinguishing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115809601A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117113162A (en) * | 2023-05-23 | 2023-11-24 | 南华大学 | Eddar-rock structure background discrimination and graphic method integrating machine learning |
| CN117332240A (en) * | 2023-12-01 | 2024-01-02 | 中铁四局集团有限公司 | Rock burst prediction model construction method, storage medium, rock burst prediction method and system |
-
2022
- 2022-12-06 CN CN202211557450.1A patent/CN115809601A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117113162A (en) * | 2023-05-23 | 2023-11-24 | 南华大学 | Eddar-rock structure background discrimination and graphic method integrating machine learning |
| CN117113162B (en) * | 2023-05-23 | 2024-02-02 | 南华大学 | Eddar-rock structure background discrimination and graphic method integrating machine learning |
| CN117332240A (en) * | 2023-12-01 | 2024-01-02 | 中铁四局集团有限公司 | Rock burst prediction model construction method, storage medium, rock burst prediction method and system |
| CN117332240B (en) * | 2023-12-01 | 2024-04-16 | 中铁四局集团有限公司 | Rock burst prediction model construction method, storage medium, rock burst prediction method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111783825B (en) | Logging lithology recognition method based on convolutional neural network learning | |
| CN115809601A (en) | Sedimentary rock structure background distinguishing method | |
| WO2019149147A1 (en) | Method for dividing ecological and geological environment types based on coal resource development | |
| CN112989708B (en) | Well logging lithology identification method and system based on LSTM neural network | |
| CN110232280B (en) | Software security vulnerability detection method based on tree structure convolutional neural network | |
| CN113297174B (en) | Land utilization change simulation method based on deep learning | |
| CN113344050A (en) | Lithology intelligent identification method and system based on deep learning | |
| CN112761628B (en) | Shale gas yield determination method and device based on long-term and short-term memory neural network | |
| CN114816497B (en) | Link generation method based on BERT pre-training model | |
| CN112712025A (en) | Complex lithology identification method based on long-term and short-term memory neural network | |
| CN114781538A (en) | Air quality prediction method and system of GA-BP neural network coupling decision tree | |
| CN117272841B (en) | Shale gas dessert prediction method based on hybrid neural network | |
| CN113779880A (en) | Tunnel surrounding rock two-dimensional quality evaluation method based on advanced drilling data | |
| CN110600135A (en) | Breast cancer prediction system based on improved random forest algorithm | |
| Chu et al. | Co-training based on semi-supervised ensemble classification approach for multi-label data stream | |
| CN117370766A (en) | Satellite mission planning scheme evaluation method based on deep learning | |
| CN115965020A (en) | Knowledge extraction method for wide-area geographic information knowledge graph construction | |
| CN118133104A (en) | Rapid identification method for lithofacies of deep sea-phase shale gas well | |
| CN117909873A (en) | Method and device for realizing big data processing facing data security, processor and computer readable storage medium thereof | |
| CN113284620B (en) | Establishment method of occupational health data analysis model | |
| CN113762394B (en) | Blasting block prediction method | |
| CN115660221A (en) | Oil and gas reservoir economic recoverable reserve assessment method and system based on hybrid neural network | |
| CN114818493A (en) | Method for quantitatively evaluating integrity degree of tunnel rock mass | |
| CN113642772A (en) | Logging reservoir identification and prediction method based on machine learning | |
| CN112907698A (en) | Logging curve generation method dynamically fusing time sequence and non-time sequence characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |