CN117235477B - User group evaluation method and system based on deep neural network - Google Patents
User group evaluation method and system based on deep neural network Download PDFInfo
- Publication number
- CN117235477B CN117235477B CN202311507410.0A CN202311507410A CN117235477B CN 117235477 B CN117235477 B CN 117235477B CN 202311507410 A CN202311507410 A CN 202311507410A CN 117235477 B CN117235477 B CN 117235477B
- Authority
- CN
- China
- Prior art keywords
- evaluation
- data
- user group
- exercise
- items
- 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.)
- Active
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 231
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 120
- 238000000034 method Methods 0.000 claims abstract description 62
- 238000013210 evaluation model Methods 0.000 claims abstract description 40
- 239000011159 matrix material Substances 0.000 claims abstract description 27
- 239000013598 vector Substances 0.000 claims description 39
- 238000004364 calculation method Methods 0.000 claims description 27
- 230000008569 process Effects 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 15
- 238000007781 pre-processing Methods 0.000 claims description 14
- 210000002569 neuron Anatomy 0.000 claims description 13
- 238000013507 mapping Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 230000017105 transposition Effects 0.000 claims description 4
- 238000005094 computer simulation Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 abstract description 4
- 238000011157 data evaluation Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 17
- 238000003860 storage Methods 0.000 description 11
- 238000013135 deep learning Methods 0.000 description 10
- 238000004422 calculation algorithm Methods 0.000 description 9
- 238000012360 testing method Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 5
- 238000013075 data extraction Methods 0.000 description 5
- 210000004556 brain Anatomy 0.000 description 4
- 238000004140 cleaning Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013501 data transformation Methods 0.000 description 2
- 230000007123 defense Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000008485 antagonism Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the technical field of data evaluation, and provides a user group evaluation method and system based on a deep neural network. The method comprises the following steps: establishing an evaluation data dictionary based on exercise data samples of the historical user group; establishing a training data set according to the evaluation data dictionary and the evaluation results of the historical user group; based on the deep neural network, constructing a user group evaluation model, and training the user group evaluation model by using a training data set; according to the evaluation data dictionary, determining a plurality of evaluation parameters of the user group to be evaluated, and generating an input matrix corresponding to each evaluation parameter; and inputting an input matrix of the user group to be evaluated into a user group evaluation model to evaluate the comprehensive countermeasure capability of the user group to be evaluated. The invention avoids the limitation of single node data, reduces the cost of exercise training, saves the time of exercise training, and obtains the optimization effect based on exercise training task requirements.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a user group evaluation method and system based on a deep neural network.
Background
Team resistance is the ability or ability of a mission team to achieve a defensive objective under specific constraints. The inherent capability of the counter force as a team is generally not changed with the specific exercise training process and the dynamic evolution of the exercise training situation, but rather, the counter force is determined by relative static elements such as system programming, exercise training, defense deployment and the like of the team. There is a division of strength in the opposition force, the scale of which is called the opposition force index, which is an accurate representation of the opposition force in real space, by means of which a quantitative calculation of the team's opposition force can be achieved. Therefore, the method has important application value in the fields of defense scheme planning, countermeasure structure optimization, military theory research and the like. Currently, there are classical methods for calculating the team's counter force index, such as methods of Dupi index, dennet index, power index, etc., all of which adopt the idea of a reduction theory, and the team's counter force index is obtained by linear summation of each component element of the team and the counter force index. The existing method is simple and visual, but does not embody the complex nature of informatization, and especially ignores the important influence of the opposing force in the system training process. In addition, how to accurately and rapidly evaluate team counter force of system training is a problem to be solved at present.
However, a large amount of raw data is accumulated in each exercise training of each training base, most of the data are used for real-time situation display, most of the data use data structures and data sets defined by the base, the data structures are difficult to unify, post-processing and analysis are lacked, the data are difficult to support predictive evaluation of exercise training tasks, the predictive evaluation cannot objectively and accurately reflect actual countermeasure capability of a team without data support, exercise training cost is increased, and the method also becomes a bottleneck for improving defending capability of the team. In addition, there is still a great room for improvement in various aspects such as application of data related to the countermeasure ability or defensive ability of the exercise training task, and accuracy of prediction of the countermeasure ability or defensive ability.
Therefore, it is necessary to provide a user group evaluation method based on a deep neural network to solve the above-mentioned problems.
Disclosure of Invention
The invention aims to provide a user group evaluation method and a system based on a deep neural network, which are used for solving the technical problems that in the prior art, the data structure of each exercise training related data of each training base is diversified and not easy to unify, the data is difficult to be used for the predictive evaluation of exercise training tasks due to the lack of post-processing and analysis of the data, the actual countermeasure capability or defensive capability of each team cannot be objectively and accurately evaluated by the existing method, the exercise training cost is increased, and the like.
The first aspect of the present invention provides a user group evaluation method based on a deep neural network, including: establishing an evaluation data dictionary based on exercise data samples of the historical user group; establishing a training data set according to the evaluation data dictionary and the evaluation results of the historical user group; based on a deep neural network, constructing a user group evaluation model, and training the user group evaluation model by using the training data set; according to the evaluation data dictionary, determining a plurality of evaluation parameters of a user group to be evaluated, and generating an input matrix corresponding to each evaluation parameter; and inputting the input matrix of the user group to be evaluated into a trained user group evaluation model, and evaluating the comprehensive countermeasure capability of the user group to be evaluated.
According to an alternative embodiment, building the assessment data dictionary based on exercise data samples of the historical user group comprises: and extracting characteristic data of exercise training data of historical user groups of a plurality of bases to obtain the following general data items: the method comprises the steps of performing a exercise object, performing object dividing items, performing stages, performing capability categories, performing capability multi-level dividing items, performing targets, performing target completion conditions, evaluating items, evaluating sub-items, evaluating indexes and evaluating points; and establishing an evaluation data dictionary based on the universal data item.
According to an alternative embodiment, preprocessing is carried out on each general data item, and a multi-level evaluation vector corresponding to each general data item is determined to obtain a data set corresponding to each general data item; the preprocessing includes converting qualitative data into quantitative data using a computational model.
According to an alternative embodiment, one or more general data items are selected from the general data items as evaluation parameters, and the evaluation result of each historical user group is determined according to the multi-level evaluation vector of each general data item, so as to establish a training data set.
According to an alternative embodiment, the training the user group assessment model using the training dataset comprises: and vector conversion is carried out on all general data items of each group of data in the evaluation data dictionary, the general data items are used as neuron input of a deep neural network, comprehensive countermeasure results of all historical user groups are used as model output, and a mapping relation between all general data items and evaluation results of all the historical user groups is obtained.
According to an alternative embodiment, according to the preset training times and the preset model accuracy, judging whether the actual model output in the training process is consistent with the expected model output or not so as to determine that the model training of the user group evaluation model is completed.
According to an alternative embodiment, the method determines a plurality of evaluation parameters of the user group to be evaluated according to the evaluation data dictionary, and generates an input matrix corresponding to each evaluation parameter: according to the identification information of the user group to be evaluated, selecting exercise objects, exercise stages, exercise targets, exercise capacity, evaluation items, evaluation indexes and evaluation points from the evaluation data dictionary as evaluation parameters, respectively converting the evaluation parameters into respective corresponding vectors, and obtaining an input matrix of the user group to be evaluated after normalization and transposition operation, wherein each row in the input matrix represents neurons corresponding to one general data item, and each column represents a group of data corresponding to each general data item.
According to an alternative embodiment, the input matrix of the user group to be evaluated is input into a trained user group evaluation model, and the antagonism capability evaluation value of the user group to be evaluated is output.
The second aspect of the present invention proposes a user group evaluation system, which adopts the user group evaluation method of the first aspect of the present invention to predict the comprehensive countermeasure capability of a user group, including an interactable server and a user; the user terminal is used for selecting the evaluation parameters of the user group to be evaluated from the evaluation data dictionary by a user and inputting the evaluation parameters into an input interface; the server comprises a calling module, a user group evaluation model is called based on the input of the evaluation parameters, a prediction result is output, and the prediction result is returned to the user.
According to an optional implementation manner, the prediction result and the prediction calculation process are displayed on a display page of the user side in a chart and text mode, wherein feature data extraction is performed on exercise training data of historical user groups of a plurality of bases, and the following general data items are obtained: the method comprises the steps of performing a exercise object, performing object dividing items, performing stages, performing capability categories, performing capability multi-level dividing items, performing targets, performing target completion conditions, evaluating items, evaluating sub-items, evaluating indexes and evaluating points; and establishing an evaluation data dictionary based on the universal data item.
A third aspect of the present invention provides an electronic apparatus, comprising: one or more processors; a storage means for storing one or more programs; the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of the first aspect of the present invention.
A fourth aspect of the invention provides a computer readable medium having stored thereon a computer program which, when executed by a processor, implements the method according to the first aspect of the invention.
The embodiment of the invention has the following advantages:
According to the invention, the exercise training data is taken as a core, redundant data is removed, a core evaluation index is found out, an evaluation data dictionary is established, a training data set is established, a user group evaluation model is established in combination with deep learning to provide a prediction evaluation service, the exercise training is predicted and evaluated from the whole environment, the comprehensive countermeasure capability of the user group to be evaluated can be accurately predicted, the limitation of single node data can be effectively avoided, the cost of the exercise training can be reduced, the time of the exercise training can be saved, the optimization effect based on the exercise training task requirement can be obtained, and the exercise time cost and the repeatability can be effectively reduced.
Drawings
FIG. 1 is a flow chart of steps of an example of a deep neural network based user group evaluation method of the present invention;
FIG. 2 is an exemplary diagram of a deep neural network architecture of a user group evaluation model of the present invention;
FIG. 3 is a schematic diagram of an example of a user group evaluation system according to the present invention;
FIG. 4 is a schematic diagram of the functional composition of a user group assessment system according to the present invention;
FIG. 5 is a schematic diagram of an embodiment of an electronic device according to the present invention;
fig. 6 is a schematic diagram of an embodiment of a computer readable medium according to the present invention.
Detailed Description
It should be noted that, in the case of no conflict, the embodiments and features in the embodiments may be combined with each other. The invention will be described in detail below with reference to the drawings in connection with embodiments.
In view of the above problems, the present invention provides a deep neural network-based user group evaluation method, which takes exercise training data as a core, eliminates redundant data, and finds out a core evaluation index. By establishing an evaluation data dictionary, establishing a training data set, combining deep learning to construct a user group evaluation model to provide a prediction evaluation service, predicting and evaluating exercise training from the overall environment, the comprehensive countermeasure capability of the user group to be evaluated can be accurately predicted, the limitation of single node data can be effectively avoided, the cost of exercise training can be reduced, the time of exercise training can be saved, the optimization effect based on exercise training task requirements can be realized, and the exercise time cost and repeatability can be effectively reduced.
It is noted that Neural Networks (NNs) are a subset of ML in which computer systems are designed to work like the human brain by classifying information. Deep Learning (DL) is a subset of ML, and uses multi-layer artificial neural networks to solve complex problems such as object detection, speech recognition, and language translation. ML algorithms can be categorized into supervised, unsupervised and reinforcement learning. ML and DL differ in the way features are extracted. Traditional ML methods require the data worker to explicitly extract features by applying learning algorithms. On the other hand, in the case of DL, these characteristics are automatically learned by an algorithm, and no characteristic engineering is required. This is an advantage of DL over traditional ML methods, but in terms of team counter force predictions for system or community counter force. Deep learning is based on artificial neural networks, which are inspired by brain neurons and are designed to identify patterns in complex data. Whereas neurons in the brain are organized into billions of vast networks, each neuron is typically connected to thousands of other neurons, typically in a continuous layer, particularly in the cortex (i.e., outer brain layer). The neural network has input, output and hidden layers. A neural network with two or more hidden layers is called a deep neural network. The benefits of deep learning are used in this application to train reliable predictive models. The following will specifically explain the content of the present invention.
Example 1
Fig. 1 is a flowchart illustrating steps of an exemplary deep neural network-based user group evaluation method of the present invention.
The following describes the present invention in detail with reference to fig. 1 and 2.
First, in step S101, an evaluation data dictionary is built based on exercise data samples of a history user group.
Specifically, exercise training data of historical user groups of a plurality of bases is acquired. The plurality of bases includes a plurality of army bases, a plurality of air force bases, a plurality of navy bases, and the like.
More specifically, the exercise training data of each training base includes a large amount of raw data including qualitative data and quantitative data. The qualitative data includes the following data: words representing names of units in exercise training (for example, "command post", "countermeasure unit", etc.), classification data (data obtained by classifying according to exercise criteria, for example, "user category", "countermeasure type", "equipment type for countermeasure", etc.), sequence data (data obtained by arranging according to exercise criteria sequence, for example, "score level", "personnel level", etc.), attribute data (data describing attributes of units, for example, "color", "shape", "size", etc.), attitude data (data of mindsets, ideas or trends of things or phenomena, for example, "satisfaction", "trust", "support", etc. judgment). The quantitative data is data which can be represented by specific numerical values, can be subjected to mathematical operation and analysis, and can reflect the quantity characteristics and the quantity relation of things, such as 'capability score', 'marching path length', 'task time consumption', and the like.
In this example, qualitative data and quantitative data are relative, the quantitative data being represented in a numerical form, and the qualitative data being represented in a non-numerical form.
In this embodiment, exercise training data of a historical user group of a plurality of bases, for example, a plurality of army bases, is preprocessed, and feature data extraction is performed to obtain a general data item.
The preprocessing comprises redundant data elimination, data cleaning and the like, and characteristic data extraction is carried out, namely core evaluation indexes are found out, so that universal data items are obtained. The method specifically comprises the following data processing steps:
and (3) data cleaning: inputting an original data file, mapping the original data file with a data template in a data cleaning module, deleting non-mapping data according to a mapping result, filling a missing value, processing an abnormal value, deleting repeated data and the like, and cleaning the data;
data transformation processing: the data are normalized, standardized, normalized and the like, so that the comparability among different features is realized, and the convergence rate of the model is increased;
and (3) data integration processing: combining a plurality of data sources after data transformation processing into a unified data set based on a weight integration strategy, and dividing the data set into a training set and a testing set for training the performance of a model and verifying the performance of the model;
And (3) feature selection processing: and extracting common indexes from the historical exercise training data according to each exercise training task to form core evaluation indexes, wherein the core evaluation index data are selected from the original data to obtain general data items.
Specifically, the general data items comprise exercise objects, exercise object dividing items, exercise stages, exercise capabilities, exercise capability categories, exercise capability multi-level dividing items, exercise targets, exercise target completion conditions, evaluation items, evaluation sub-items, evaluation indexes and evaluation points. The exercise object refers to a user group (i.e., a team with a certain number of users) participating in exercise training (e.g., exercise training), the user groups may be divided according to the number of users, for example, the user group with the number of users less than a first designated number (e.g., less than 100 people) is a first user group, the user group with the number of users greater than or equal to the first designated number and less than a second designated number (e.g., greater than 1000 people) is a second user group, the user group with the number of users greater than or equal to the second designated number is a third user group, and so on.
Further, the score of each user group in the process of exercise training or competition fighting training is also included.
The scores of each user group in the process of exercise training or competition fighting training are obtained by the following modes:
for quantitative data, each quantitative data item node in the general data item is provided with a 1-to-1 configuration calculation model, calculation is carried out according to the standard of the data item through the collected quantitative data, and calculation scores are returned;
for qualitative data, firstly converting the qualitative data into quantitative data according to exercise standards, and then obtaining scores by multiplexing the calculation mode of the quantitative data;
it should be noted that, the exercise stage refers to a general stage of exercise training, and there are multiple stages in exercise training, such as an countermeasure stage, a remote delivery stage, and a countermeasure execution stage, where each stage has a score, and typically the score of each stage is not a single value, for example, is a multidimensional vector. The exercise capability refers to various capabilities which are shown in one exercise training, and is also important data for measuring exercise results, such as scout information capability, fire striking capability, multidimensional protection capability, comprehensive guarantee capability and the like, and the exercise capability is represented by vectors, such as multidimensional vectors. The evaluation item refers to a specific evaluation classification in exercise training, for example, a exercise target is split into a plurality of items (evaluation sub-items), and the completion condition of each item is evaluated and expressed in a vector form, for example, as a multidimensional vector. The evaluation index refers to the completion condition of each index in exercise training, such as target discovery and identification condition, information access condition, countermeasure conversion, enemy fire prevention condition and the like.
The evaluation points are used for processing the input exercise data to form N acquisition items, comprehensively evaluating the input data according to rules, and calculating scores, wherein the scores are expressed in a vector form, such as multidimensional vectors.
The evaluation point is a specific evaluation element in the exercise. For example, intelligence reconnaissance capability may include multiple evaluation points such as discovery rate, identification rate, positioning accuracy, time of generation target, etc. Each evaluation point has specific standards and requirements, which are evaluation standards formed by users and experts in the exercise site before exercise and are used for measuring the actual situation of the evaluated object. The selection of the evaluation point depends on the evaluation purpose and the characteristics of the evaluation object, and flexible adjustment is required according to the actual situation.
And establishing an evaluation data dictionary based on the universal data item. Specifically, an attribute name, a data type (for example, a single value) and a score calculation method (positive and negative samples corresponding to each general data item are determined) will be established for each general data item. For example, the exercise object is a single value, the exercise object with the score of 90 or more is a positive sample, and the exercise object with the score of less than 90 is a negative sample. For example, in the exercise stage, vector representation is used, and vector aggregation operation is required to be performed on vectors, so that the vectors are converted into single values, for example: the exercise stage with the single value greater than or equal to 90 minutes is taken as a positive sample, and the exercise stage with the single value less than 90 minutes is taken as a negative sample, and the specific reference can be seen in the following table 1.
TABLE 1
Specifically, preprocessing is performed on each general data item, a multi-level evaluation vector corresponding to each general data item is determined, and a data set corresponding to each general data item is obtained, wherein the preprocessing comprises the step of converting qualitative data into quantitative data by using a calculation model.
Preprocessing each general data item, for example, configuring a calculation model, and because qualitative data exists in exercise training data, manually defining rules on the qualitative data by configuring the calculation model, and converting the qualitative data into quantitative data.
In one embodiment, for the pretreatment of exercise object data, specifically, the task completion situation of the exercise object is extracted from the related database, weighted average is performed on each task completion situation, and the score of the exercise object is obtained, for example, the range is from 0 to 100, the exercise object is divided into four classes, 0 to 59 is a failed class, 60 to 69 is a failed class, 70 to 89 is good, and 90 to 100 is excellent. Differentiation of exercise subjects is divided into four stages, such as a first stage, a second stage, a third stage, and a fourth stage.
For the data preprocessing of the exercise stage, the exercise stage is obtained from a related database, the completion condition of each stage is weighted and averaged, the score condition of each stage is calculated, the score range of each stage is divided from 0 to 100 by four stages, 0 to 59 is a failed grade, 60 to 69 is a good grade, 70 to 89 is good, and 90 to 100 is excellent.
For the exercise capability data preprocessing, the exercise capability is extracted from a related database, if quantitative data exist, the exercise capability can be directly used, if qualitative data exist, the scores of the general capabilities such as the reconnaissance information capability, the fire striking capability, the multidimensional protection capability, the comprehensive guarantee capability and the like are calculated through a calculation model, each capability score ranges from 0 to 100, the four-level system is divided, 0 to 59 is a failed grid, 60 to 69 is a failed grid, 70 to 89 is good, and 90 to 100 is excellent.
For the pretreatment of the evaluation item data, the evaluation item is split into a plurality of targets, the targets are stored in a quantitative data form, the system automatically finishes scoring, the final score is extracted during pretreatment, each evaluation item score ranges from 0 to 100, the evaluation item is divided into four grades, 0 to 59 is a failing grade, 60 to 69 is a passing grade, 70 to 89 is good, and 90 to 100 is excellent.
And preprocessing evaluation index data, specifically extracting general evaluation indexes such as target discovery identification conditions, information access conditions, countermeasure conversion, protection firepower conditions and the like from a database, and obtaining the completion condition of each index. If quantitative data is available for direct use, and if only qualitative data is available, the score of each capability is calculated by a calculation model, the score of each evaluation index ranges from 0 to 100, is divided into four levels, 0 to 59 is failed, 60 to 69 is failed, 70 to 89 is good, and 90 to 100 is excellent.
For the preprocessing of evaluation point data, specific evaluation elements in exercise training are processed, an index system is formed by the input exercise data, evaluation points are split into a plurality of data items according to the index system, evaluation calculation is carried out on the data items through a calculation model, a calculation result is converted into an evaluation point score vector, the score range of each evaluation point is divided from 0 to 100 by four levels, 0 to 59 is a failed grid, 60 to 69 is a good grid, 70 to 89 is a good grid, and 90 to 100 is a good grid.
Further, a multi-level evaluation vector corresponding to each general data item is obtained, for example, a single-dimensional evaluation vector, a multi-dimensional evaluation vector or a specified dimension vector is obtained, so as to obtain a data set corresponding to each general data item, namely, one general data item corresponds to one data set.
Specifically, according to the configured calculation model, the evaluation vector of each general data item is obtained to form a data set, and the data set is 5050 rows and 6 columns, for example. Thereby creating an evaluation data dictionary comprising a plurality of generic data items, a multi-level evaluation vector corresponding to each generic data item, and a data set corresponding to each generic data item. Wherein the data set contains exercise training data from a plurality of exercise base exercises, which can be used to predict exercise training results.
In an embodiment, the evaluation data dictionary is used for querying general data items of the user group, multi-level evaluation vectors corresponding to the general data items, and the like.
In another embodiment, in the subsequent deep neural network, the multi-level evaluation vectors corresponding to the general data items are all used as input for linear increase. The format of the above evaluation data dictionary is extracted into the memory in a segmented manner, and then the exercise object, exercise stage, exercise capability, evaluation item, evaluation index, evaluation point data are saved into Comma Separated value (CSV, sometimes referred to as character Separated value) file. Because the csv file can only store character type data, the floating point number list can only be stored in the csv in the form of character strings, and the character strings are converted into the list again during reading.
In this example, one general data item corresponds to one data set, but the present invention is not limited thereto, and in other examples, one general data item may correspond to a plurality of data sets. In addition, in view of versatility, simplicity, scalability, and compatibility, the various items of use and the various features are stored in the CSV file, so that data exchange is easier and further data processing and analysis is easier. The foregoing is illustrative only and is not to be construed as limiting the invention.
Next, in step S102, a training data set is established according to the evaluation data dictionary and the evaluation results of the historical user group.
And establishing a training data set according to all general data items in the evaluation data dictionary, the data set corresponding to all general data items and the evaluation result of the historical user group, wherein the training data set is used for training a user group evaluation model.
Specifically, one or more general data items are selected from the general data items to serve as evaluation parameters, and evaluation results of each historical user group are determined according to multi-level evaluation vectors of the general data items so as to establish a training data set. Wherein the evaluation result is extracted, for example, by directed analysis of a relational database.
The relation database is a exercise evaluation data storage database, and the relation database comprises a plurality of types of data such as an evaluation index system, evaluation task information, original data, calculation process data, calculation result data, evaluation result data and the like, wherein the calculation result data is calculated according to exercise training data, and the evaluation result is manually marked according to the calculation result data.
In the present embodiment, six general data items including a exercise object, an exercise stage, exercise ability, an evaluation item, an evaluation index, and an evaluation point are selected as evaluation parameters, but not limited thereto, and in other examples, the general data items may further include an exercise ability category, an exercise ability multistage division item, an exercise target, a completion condition of the exercise target, and the like, which are described only as optional examples and are not to be construed as limiting the present invention.
Optionally, a test dataset is built from the evaluation data dictionary, for example comprising 1382 rows and 6 columns of data, for verifying model prediction accuracy of the user group evaluation model.
It should be noted that the foregoing is merely illustrative of the present invention and is not to be construed as limiting thereof.
Next, in step S103, a user group evaluation model is constructed based on the deep neural network, the user group evaluation model being trained using the training data set.
Specifically, the method for setting the parameters of the neural network specifically comprises the following steps: training times, training target, learning rate, momentum factor and interval times, wherein the training times determine the total cycle number of the model training process and also influence the model training time. In this embodiment, the training number is, for example, not less than 2000. The learning rate, momentum factor, and number of intervals are selected according to the size of the dataset.
Further, determining a training objective includes defining an accuracy of the user group assessment model, e.g., 85% -98%, for determining whether the actual model output and the expected model output agree to determine completion of the model training process.
Then, a deep neural network is constructed to construct a user group evaluation model, which specifically comprises a definition input layer and an output layer, wherein the user group evaluation model takes a specified dimension vector converted by each general data item as model input, takes comprehensive countermeasure score (namely comprehensive countermeasure evaluation value) as model output, and in a model construction process (specifically a model training process), the number of input layer nodes is 6, and the number of output layer nodes is 1.
The method specifically further comprises defining a hidden layer, wherein the hidden layer is a neural network with three layers of multiple inputs and single outputs, and the number of neurons of the hidden layer is 6 according to actual conditions, and specifically, see fig. 2.
The method further comprises the step of defining an excitation function, and selecting an s-type tangent function as the excitation function of the hidden layer neurons.
For the model training process, a variable learning rate momentum gradient descent algorithm with the fastest convergence is selected.
In neural networks, the variable learning rate momentum gradient descent algorithm is mainly used to optimize the parameters of the model to minimize the loss function. The specific using method is as follows:
step S201, initializing parameters.
Firstly, parameters such as weights, biases and the like of the neural network are initialized.
Step S202, calculating gradient.
For each training sample or batch of training samples, a gradient of the loss function for each parameter is calculated from the forward propagation result of the neural network and the real labels.
And step S203, updating parameters.
The value of each parameter is updated in the opposite direction of the gradient using a gradient descent algorithm. The updated formula is as follows:
new parameter = original parameter-learning rate gradient + momentum term last updated parameter (1).
Wherein, the learning rate is a super parameter, namely controlling the step length of each update; the dynamic term is used for representing the coefficient for accelerating the convergence speed and reducing the oscillation;
it should be noted that, in the update formula of this gradient descent algorithm: "new parameter" refers to the updated parameter value. This value is calculated from the original parameters, learning rate, gradient and momentum terms; "original parameters" refer to parameter values before updating, i.e., initial values of parameters in the current iteration step; "gradient" is the partial derivative of the loss function (or objective function) with respect to the parameter, which represents the rate of change of the function value in parameter space, indicating the direction in which the function value increases fastest. In the gradient descent algorithm, updates are made in the opposite direction of the gradient in order to expect the minimum point of the function to be reached.
Assume that a quadratic loss function J (θ) = (θ -a)/(2) is used, where θ is a parameter to be optimized and a is a target value. The goal is to find a set of values for θ such that J is minimal.
Assuming that the initial parameter θ_0=5, the learning rate η=0.1, the momentum factor β=0.9, and the target value a=10.
First, the gradient of the loss function under the initial parameters is calculated: ∇ J (θ_0) =2 (θ_0-a) =2 (5-10) = -10
Then, the value of the parameter is updated using the update formula (1) of the gradient descent algorithm: θ1=θ_0- η ∇ J (θ_0) +β (θ_0- θ_ (-1))=5-0.1 (-10) +0.9 (5- θ_ (-1))=5+1+0.9 (5- θ_ (-1))=6+4.5-0.9θ_ (-1) =10.5-0.9θ_ (-1)
Wherein θ_1 represents the updated parameter value; θ_ (-1) represents the parameter value of the last iteration, since this is the first iteration, no value of the last iteration is available, which can be set to 0 or an initial value, so that the parameter value θ_1 after the first iteration is obtained, and the iteration is continued, updating the value of the parameter each time until the loss function converges to a minimum value; η represents a learning rate; beta represents a momentum factor; a represents a target value.
Step S204, repeatedly executing: steps S202 and S203 are repeated until a preset number of iterations is reached or other stop conditions are met.
Specifically, the training data set established in the step S102 is used to train the user group evaluation model, wherein the general data items of each group of data in the evaluation data dictionary are vector-converted and then used as neuron inputs of the deep neural network, the evaluation results of each user group, namely the comprehensive countermeasure score of each user group, are used as model outputs to normalize, so as to convert the evaluation results into values of 0% -100%, and the mapping relation between the general data items and the evaluation results of each historical user group is obtained.
Further, when model training is finished, testing the user group evaluation model by using a test sample in a test data set, and when a test result meets a training target (for example, model accuracy is 88%), judging that the neural network construction of the user group evaluation model is finished, in other words, finishing model training. And when the test result does not meet the training target (for example, the model precision is 88%), continuing to train the user group evaluation model until the test result meets the training target (for example, the model precision is 88%), and ending the model training process to obtain the trained user group evaluation model.
In another embodiment, according to a preset training frequency (for example, less than 2000 times) and a preset model accuracy (for example, model accuracy is 90%), whether the actual output and the expected output of the model in the training process are consistent is judged, so as to determine that model training of the user group evaluation model is completed, and a trained user group evaluation model is obtained.
It should be noted that the foregoing is merely illustrative of the present invention and is not to be construed as limiting thereof.
Next, in step S1104, a plurality of evaluation parameters of the user group to be evaluated are determined according to the evaluation data dictionary, and an input matrix corresponding to each evaluation parameter is generated.
Specifically, according to the identification information of the user group to be evaluated, selecting exercise objects, exercise stages, exercise targets, exercise capabilities, evaluation items, evaluation indexes and evaluation points from the evaluation data dictionary as evaluation parameters, respectively converting the evaluation parameters into respective corresponding vectors, and obtaining an input matrix of the user group to be evaluated after normalization and transposition operations, wherein the input matrix is a two-dimensional matrix, each row in the matrix represents neurons corresponding to one general data item, and each column represents a group of data corresponding to each general data item.
For example, the input matrix of the user group to be evaluated is the following matrix:
。
more specifically, the identification information includes code information such as a number of the user group, a two-dimensional code, etc., for identifying the user group, by which data can be queried from a related database or an evaluation data dictionary.
In an alternative embodiment, the identification information is associated with feature information of the user group, such as a historical exercise training feature, the number of users of the user group, and the like. Through the identification information of the user group to be evaluated, a similar user group similar to the user group to be evaluated can be queried, and the evaluation parameters of the similar user group are used as the evaluation parameters of the user group to be evaluated, so as to calculate the respective corresponding vectors, and further calculate the input matrix of the user group to be evaluated.
It should be noted that the foregoing is merely illustrative of the present invention and is not to be construed as limiting thereof.
In step S105, the input matrix of the user group to be evaluated is input into a trained user group evaluation model, and the comprehensive countermeasure capability of the user group to be evaluated is evaluated.
In one embodiment, the input matrix of the user group to be evaluated obtained in step S104 is input into a trained user group evaluation model, and a challenge ability evaluation value of the user group to be evaluated is output, where the challenge ability evaluation value is a score between 0 and 100, for example.
In another embodiment, the input matrix of the user group to be evaluated obtained in step S104 is input into a trained user group evaluation model, and the challenge capability level (or defensive capability level) and the score of the user group to be evaluated are output, where the challenge capability level (or defensive capability level) is classified based on the score, for example, more than 90 is classified into a first level, less than or equal to 90 is classified into a second level, and less than or equal to 80 is classified into a third level.
It should be noted that the foregoing is merely illustrative of the present invention and is not to be construed as limiting thereof. Furthermore, the drawings are only schematic illustrations of processes involved in a method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It will be readily understood that the processes shown in the figures do not indicate or limit the temporal order of these processes. In addition, it is also readily understood that these processes may be performed synchronously or asynchronously, for example, among a plurality of modules.
Compared with the prior art, the invention takes exercise training data as a core, eliminates redundant data, finds out core evaluation indexes, establishes an evaluation data dictionary, establishes a training data set, establishes a user group evaluation model in combination with deep learning to provide prediction evaluation service, predicts and evaluates exercise training from the whole environment, can accurately predict comprehensive countermeasure capability of a user group to be evaluated, can effectively avoid limitation of single node data, can reduce exercise training cost, can save exercise training time, obtains optimization effect based on exercise training task requirements, and can effectively reduce exercise time cost and repeatability.
Example 2
The following are system embodiments of the present invention that may be used to perform method embodiments of the present invention. For details not disclosed in the system embodiments of the present invention, please refer to the method embodiments of the present invention.
Fig. 3 is a schematic diagram of an example of a user group evaluation system according to the present invention.
Referring to fig. 3, a second aspect of the present disclosure provides a user group evaluation system, which performs comprehensive countermeasure capability prediction for a user group by using the user group evaluation method described in the first aspect of the present invention. The user group evaluation system comprises an interactable server side and a user side.
Fig. 4 is a schematic structural diagram of functional components of the user group evaluation system according to the present invention.
As shown in fig. 4, the main functions of the user group evaluation system are composed of the following functions: evaluation index system management, evaluation task management, evaluation data acquisition, evaluation calculation and evaluation result management.
Specifically, the user side is used for a user to select an evaluation parameter of a user group to be evaluated from the evaluation data dictionary, and the evaluation parameter is input in an input interface. The server side comprises a calling module which is used for calling a user group evaluation model based on the input of the evaluation parameters, outputting a prediction result and returning the prediction result to the user side.
Further, displaying the prediction result and the prediction calculation process on a display page of the user side in a chart and text mode, wherein feature data extraction is performed on exercise training data of historical user groups of a plurality of bases, and the following general data items are obtained: the method comprises the steps of exercise objects, exercise object dividing items, exercise stages, exercise capability categories, exercise capability multi-level dividing items, exercise targets, exercise target completion conditions, evaluation items, evaluation sub-items, evaluation indexes and evaluation points.
In one embodiment, feature data extraction is performed on exercise training data of historical user groups of a plurality of bases to obtain the following general data items: the method comprises the steps of exercise objects, exercise object dividing items, exercise stages, exercise capability categories, exercise capability multi-level dividing items, exercise targets, exercise target completion conditions, evaluation items, evaluation sub-items, evaluation indexes and evaluation points.
Preprocessing each general data item, determining a multi-level evaluation vector corresponding to each general data item, and obtaining a data set corresponding to each general data item; the preprocessing includes converting qualitative data into quantitative data using a computational model.
Further, an evaluation data dictionary is established based on the generic data items.
It should be noted that, in the system embodiment of the present invention, the establishment of the evaluation data dictionary and the establishment of the user group evaluation model are substantially the same as those in the method embodiment of the present invention, and therefore, the description of the same parts is omitted.
The user group evaluation system comprises a user group prediction service, wherein the user group prediction service is an application system matched with an evaluation prediction model for use, specifically, before the exercise training is started, a user selects exercise objects, exercise stages, exercise targets, exercise capacity, evaluation items, evaluation indexes and evaluation points from an index system in a visual mode, and a background server side invokes the user group evaluation prediction model to conduct comprehensive countermeasure capacity prediction, and returns a prediction result and a prediction process to the user side in a chart and text mode.
In addition, the user group evaluation system adopts a B/S architecture, which is convenient to be deployed on a server, does not need downloading and installation of system update, and is easy to be domesticated. The background server selects, for example, a Springs frame (Springs boot), which can quickly create Spring items running independently and integrate with a main stream frame, and an embedded server program container is used, so that the application does not need to be packed into a compression package.
Optionally, the front end framework of the user end selects, for example, ant design, which has excellent feedback components, enterprise-level website appearance, accords with ARIA standards, provides functions of keyboard processing, tabulation and the like, and saves a great amount of development time in a prototype manufacturing mode.
Compared with the prior art, the system provided by the invention adopts a B/S architecture, can be convenient for server deployment, does not need downloading and installation for system updating, is easy for localization, can quickly create Spring items which independently run and integrate with a main stream framework by selecting a Springs boot (Springs boot) for example by a background server, and can realize that applications do not need to be packed into compression packages by using an embedded server program container. The front end framework of the user end selects the anti, so that the method has excellent feedback components, enterprise-level website appearance, meets ARIA standards, provides functions of keyboard processing, tabulation and the like, and can save a great amount of development time in a prototype manufacturing mode.
Example 3
Fig. 5 is a schematic structural view of an embodiment of an electronic device according to the present invention.
As shown in fig. 5, the electronic device is in the form of a general purpose computing device. The processor may be one or a plurality of processors and work cooperatively. The invention does not exclude that the distributed processing is performed, i.e. the processor may be distributed among different physical devices. The electronic device of the present invention is not limited to a single entity, but may be a sum of a plurality of entity devices.
The memory stores a computer executable program, typically machine readable code. The computer readable program may be executable by the processor to enable an electronic device to perform the method, or at least some of the steps of the method, of the present invention.
The memory includes volatile memory, such as Random Access Memory (RAM) and/or cache memory, and may be non-volatile memory, such as Read Only Memory (ROM).
Optionally, in this embodiment, the electronic device further includes an I/O interface, which is used for exchanging data between the electronic device and an external device. The I/O interface may be a bus representing one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.
It should be understood that the electronic device shown in fig. 5 is only one example of the present invention, and the electronic device of the present invention may further include elements or components not shown in the above examples. For example, some electronic devices further include a display unit such as a display screen, and some electronic devices further include a man-machine interaction element such as a button, a keyboard, and the like. The electronic device may be considered as covered by the invention as long as the electronic device is capable of executing a computer readable program in a memory for carrying out the method or at least part of the steps of the method.
From the above description of embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or may be implemented in software in combination with the necessary hardware. Thus, as shown in fig. 6, the technical solution according to the embodiment of the present invention may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a U-disk, a mobile hard disk, etc.) or on a network, and includes several commands to cause a computing device (may be a personal computer, a server, or a network device, etc.) to perform the above-described method according to the embodiment of the present invention.
The software product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The computer readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, with readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. The readable storage medium can also be any readable medium that can communicate, propagate, or transport the program for use by or in connection with the command execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).
The computer-readable medium carries one or more programs, which when executed by one of the devices, cause the computer-readable medium to implement the data interaction methods of the present disclosure.
Those skilled in the art will appreciate that the modules may be distributed throughout several devices as described in the embodiments, and that corresponding variations may be implemented in one or more devices that are unique to the embodiments. The modules of the above embodiments may be combined into one module, or may be further split into a plurality of sub-modules.
From the above description of embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or in combination with the necessary hardware. Thus, the technical solution according to the embodiments of the present invention may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a U-disk, a mobile hard disk, etc.) or on a network, and which includes several commands to cause a computing device (may be a personal computer, a server, a mobile terminal, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The exemplary embodiments of the present invention have been particularly shown and described above. It is to be understood that this invention is not limited to the precise arrangements, instrumentalities and instrumentalities described herein; on the contrary, the invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.
Claims (6)
1. A deep neural network-based user group evaluation method, comprising:
based on exercise data samples of the historical user groups, extracting feature data from exercise training data of the historical user groups of a plurality of bases to obtain the following general data items: the method comprises the steps of performing exercise objects, exercise object dividing items, exercise stages, exercise capability categories, exercise capability multi-level dividing items, exercise targets, completion conditions of the exercise targets, evaluation items, evaluation sub-items, evaluation indexes and evaluation points, wherein the evaluation points are used for processing input exercise data to form N acquisition items, comprehensively evaluating the input data according to rules, and calculating scores, and the scores are expressed in a vector form; based on the general data items, according to the configured calculation model, calculating the evaluation vector of each general data item to form a data set, thereby establishing an evaluation data dictionary; one generic data item corresponds to one data set;
According to the evaluation data dictionary and the evaluation result of the historical user group, a training data set is established, which comprises the following steps: establishing attribute names, data types and score calculation methods for defining all the universal data items, and determining positive samples and negative samples corresponding to the universal data items;
based on a deep neural network, constructing a user group evaluation model, and training the user group evaluation model by using the training data set;
according to the evaluation data dictionary, determining a plurality of evaluation parameters of a user group to be evaluated, and generating an input matrix corresponding to each evaluation parameter, wherein the method specifically comprises the following steps: selecting exercise objects, exercise stages, exercise targets, exercise capabilities, evaluation items, evaluation indexes and evaluation points from the evaluation data dictionary as evaluation parameters according to the identification information of the user group to be evaluated, respectively converting the evaluation parameters into respective corresponding vectors, and obtaining an input matrix of the user group to be evaluated after normalization and transposition operations, wherein each row in the input matrix represents neurons corresponding to one general data item, and each column represents a group of data corresponding to each general data item;
and inputting the input matrix of the user group to be evaluated into a trained user group evaluation model, and evaluating the comprehensive countermeasure capability of the user group to be evaluated.
2. The method for deep neural network based user group evaluation of claim 1,
preprocessing each general data item, determining a multi-level evaluation vector corresponding to each general data item, and obtaining a data set corresponding to each general data item; the preprocessing includes converting qualitative data into quantitative data using a computational model.
3. The method for deep neural network based user group evaluation of claim 1,
and selecting one or more general data items from the general data items as evaluation parameters, and determining the evaluation result of each historical user group according to the multi-level evaluation vector of each general data item so as to establish a training data set.
4. The deep neural network based user group evaluation method of claim 2, wherein the training the user group evaluation model using the training dataset comprises:
and vector conversion is carried out on all general data items of each group of data in the evaluation data dictionary, the general data items are used as neuron input of a deep neural network, comprehensive countermeasure results of all historical user groups are used as model output, and a mapping relation between all general data items and evaluation results of all the historical user groups is obtained.
5. The method for deep neural network based user group evaluation of claim 1 or 4,
and judging whether the actual output of the model in the training process is consistent with the expected output of the model according to the preset training times and the preset model accuracy so as to determine that the model training of the user group evaluation model is finished.
6. A deep neural network-based user group evaluation system, which performs comprehensive countermeasure capability prediction of a user group by adopting the deep neural network-based user group evaluation method according to any one of claims 1 to 5, and is characterized by comprising an interactable server side and a user side;
the user terminal is used for selecting the evaluation parameters of the user group to be evaluated from the evaluation data dictionary by a user and inputting the evaluation parameters into the input interface; and extracting characteristic data of exercise training data of historical user groups of a plurality of bases to obtain the following general data items: the method comprises the steps of performing exercise objects, exercise object dividing items, exercise stages, exercise capability categories, exercise capability multi-level dividing items, exercise targets, completion conditions of the exercise targets, evaluation items, evaluation sub-items, evaluation indexes and evaluation points, wherein the evaluation points are used for processing input exercise data to form N acquisition items, comprehensively evaluating the input data according to rules, and calculating scores, and the scores are expressed in a vector form; based on the general data items, according to the configured calculation model, calculating the evaluation vector of each general data item to form a data set, thereby establishing an evaluation data dictionary; one generic data item corresponds to one data set; according to the evaluation data dictionary and the evaluation result of the historical user group, a training data set is established, which comprises the following steps: establishing attribute names, data types and score calculation methods for defining all the universal data items, and determining positive samples and negative samples corresponding to the universal data items; based on a deep neural network, constructing a user group evaluation model, and training the user group evaluation model by using the training data set;
The server side comprises a calling module which calls a user group evaluation model based on the input of the evaluation parameters, outputs a prediction result and returns the prediction result to the user side, wherein,
according to the evaluation data dictionary, determining a plurality of evaluation parameters of a user group to be evaluated, and generating an input matrix corresponding to each evaluation parameter, wherein the method specifically comprises the following steps: selecting exercise objects, exercise stages, exercise targets, exercise capabilities, evaluation items, evaluation indexes and evaluation points from the evaluation data dictionary as evaluation parameters according to the identification information of the user group to be evaluated, respectively converting the evaluation parameters into respective corresponding vectors, and obtaining an input matrix of the user group to be evaluated after normalization and transposition operations, wherein each row in the input matrix represents neurons corresponding to one general data item, and each column represents a group of data corresponding to each general data item;
inputting the input matrix of the user group to be evaluated into a trained user group evaluation model, and evaluating the comprehensive countermeasure capability of the user group to be evaluated; and
and displaying the prediction result and the prediction calculation process on a display page of the user side in a chart and text mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311507410.0A CN117235477B (en) | 2023-11-14 | 2023-11-14 | User group evaluation method and system based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311507410.0A CN117235477B (en) | 2023-11-14 | 2023-11-14 | User group evaluation method and system based on deep neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117235477A CN117235477A (en) | 2023-12-15 |
| CN117235477B true CN117235477B (en) | 2024-02-23 |
Family
ID=89098797
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311507410.0A Active CN117235477B (en) | 2023-11-14 | 2023-11-14 | User group evaluation method and system based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117235477B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329348A (en) * | 2020-11-06 | 2021-02-05 | 东北大学 | Intelligent decision-making method for military countermeasure game under incomplete information condition |
| CN112926739A (en) * | 2021-03-11 | 2021-06-08 | 北京计算机技术及应用研究所 | Network countermeasure effectiveness evaluation method based on neural network model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11501042B2 (en) * | 2014-03-24 | 2022-11-15 | Imagars Llc | Decisions with big data |
-
2023
- 2023-11-14 CN CN202311507410.0A patent/CN117235477B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329348A (en) * | 2020-11-06 | 2021-02-05 | 东北大学 | Intelligent decision-making method for military countermeasure game under incomplete information condition |
| CN112926739A (en) * | 2021-03-11 | 2021-06-08 | 北京计算机技术及应用研究所 | Network countermeasure effectiveness evaluation method based on neural network model |
Non-Patent Citations (3)
| Title |
|---|
| 基于大数据及机器学习的智能作战评估方法;左钦文 等;兵器装备工程学报;第41卷(第2期);第107页正文第3段，第108-109页第2节 * |
| 基于强制稀疏自编码神经网络的作战态势评估方法研究;郭圣明;贺筱媛;吴琳;胡晓峰;;***仿真学报(第03期);全文 * |
| 深度学习技术在军事领域应用;罗荣;王亮;肖玉杰;何翼;赵东峰;;指挥控制与仿真(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117235477A (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022068623A1 (en) | Model training method and related device | |
| CN106372278A (en) | Sensitivity analysis method jointly considering input parameter uncertainty and proxy model uncertainty | |
| CN112052948B (en) | Network model compression method and device, storage medium and electronic equipment | |
| CN113762486B (en) | Method and device for constructing fault diagnosis model of converter valve and computer equipment | |
| CN106650920A (en) | Prediction model based on optimized extreme learning machine (ELM) | |
| Ahmadi et al. | Learning fuzzy cognitive maps using imperialist competitive algorithm | |
| CN114548591A (en) | Time sequence data prediction method and system based on hybrid deep learning model and Stacking | |
| CN116757497A (en) | Multi-mode military intelligent auxiliary combat decision-making method based on graph-like perception transducer | |
| Czajkowski et al. | Steering the interpretability of decision trees using lasso regression-an evolutionary perspective | |
| CN112817442B (en) | FFM-based situation information classification recommendation system and method under multitasking condition | |
| Ren et al. | Modified particle swarm optimization algorithm for engineering structural optimization problem | |
| CN117371320A (en) | Photonic crystal fiber prediction method and system based on reinforced gray wolf optimization algorithm | |
| CN117235477B (en) | User group evaluation method and system based on deep neural network | |
| CN116304518A (en) | Heterogeneous graph convolution neural network model construction method and system for information recommendation | |
| Dan et al. | Application of machine learning in forecasting energy usage of building design | |
| Yuviler et al. | One pixel adversarial attacks via sketched programs | |
| CN115081856A (en) | Enterprise knowledge management performance evaluation device and method | |
| CN113191527A (en) | Prediction method and device for population prediction based on prediction model | |
| Öğütcü | Student performance score prediction using artificial neural network with the support of exploratory factor analysis and clustering | |
| CN111563767A (en) | Stock price prediction method and device | |
| WO2020100739A1 (en) | Learning device, learning method and learning program | |
| CN115620807B (en) | Method for predicting interaction strength between target protein molecule and drug molecule | |
| Deng et al. | ANNPDP: An Efficient and Stable Evaluation Engine for Large-Scale Policy Sets | |
| CN110942149B (en) | Feature variable selection method based on information change rate and condition mutual information | |
| CN112650770B (en) | MySQL parameter recommendation method based on query work load analysis |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |