Training methods, devices, media, and equipment for models of gathering places of specific personnel

Abstract

This application discloses a training method, apparatus, medium, and equipment for a model of gathering places of specific individuals, belonging to the field of machine learning technology. The method involves acquiring financial behavior characteristics, traffic behavior characteristics, and geographical region characteristics of a specific group of people; generating a first training set based on the financial and traffic behavior characteristics, the training samples including financial and traffic behavior characteristics within a first time period and a second time period, the second time period being later than the first time period; training an RNN model based on the first training set; generating a second training set based on the financial, traffic, and geographical region characteristics, the training samples including financial behavior characteristics, traffic characteristics, geographical region characteristics, and annotation information, the annotation information representing the probability that a specific individual will appear at a certain gathering place within a certain geographical region at a certain time; and training a PSO-ELM model based on the second training set to improve prediction accuracy.

Description

Technical Field

[0001] This application relates to the field of machine learning technology, and in particular to a training method, apparatus, medium and equipment for a model of a gathering place of specific people. Background Technology

[0002] Identifying the gathering places of specific individuals is crucial for public security and public health. Accurately predicting these gathering places can help law enforcement agencies more effectively monitor and combat illegal and criminal activities, while also facilitating targeted prevention and intervention measures to reduce social risks.

[0003] Traditional methods for training models of specific gathering places rely primarily on manual investigations, tip-offs, and limited data analysis. These methods suffer from inefficiency, low accuracy, and limited coverage. For example, manual investigations require significant manpower and time, and struggle to comprehensively cover all possible areas; tip-offs are uncertain and untimely; and limited data analysis may fail to fully uncover potential information and accurately predict where specific people gather. Summary of the Invention

[0004] This application provides a training method, apparatus, medium, and equipment for a model of gathering places for specific individuals, to address the problems of low prediction efficiency, low accuracy, and limited coverage of gathering places. The technical solution is as follows:

[0005] According to a first aspect of this application, a method for training a model of a gathering place for specific individuals is provided, the method comprising:

[0006] To obtain the financial behavior characteristics, transportation behavior characteristics, and geographical region characteristics of a specific population;

[0007] A first training set is generated based on the financial behavior characteristics and the traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and in a second time period, wherein the second time period is later than the first time period.

[0008] Train a recurrent neural network (RNN) model based on the first training set;

[0009] A second training set is generated based on the financial behavior characteristics, the traffic behavior characteristics, and the geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics, and annotation information of each specific person. The annotation information represents the probability that the specific person appears in a certain gathering place in a certain geographical region at a certain time.

[0010] The Particle Swarm Optimization (PSO)-Extreme Learning Machine (ELM) model is trained based on the second training set.

[0011] In one possible implementation, obtaining the financial behavior characteristics, transportation behavior characteristics, and geographical region characteristics of a specific population group includes:

[0012] The system acquires financial behavior data, traffic behavior data, and geographic area data for a specific population group. The financial behavior data includes at least one of the following: consumption location data, consumption time data, consumption amount data, and abnormal consumption frequency data. The traffic behavior data includes data on the frequency of specific individuals using public transportation in a certain functional area, passenger flow data entering and exiting transportation stations, data on the functional area to which the transportation stations belong, and correlation data between peak travel times and suspicious gathering places. The geographic area data includes at least one of the following: geographic data of each geographic area, population flow data, and historical crime data within the geographic area.

[0013] Convert the financial behavior data into financial behavior characteristics;

[0014] The traffic behavior data is converted into traffic behavior features;

[0015] The geographic region data is converted into geographic region features.

[0016] In one possible implementation, converting the financial behavior data into financial behavior features includes:

[0017] For the aforementioned consumption venue data, the code of the category to which each consumption venue belongs is used to form the first financial behavior feature;

[0018] For the consumption time data, the encoding of the time period to which each consumption time belongs is used to form a second financial behavior feature;

[0019] For the aforementioned consumption amount data, the codes of the consumption level to which each consumption amount belongs are used to form a third financial behavior feature;

[0020] For the abnormal consumption frequency data, the abnormal consumption frequency is calculated based on the number of abnormal consumptions and the total number of consumptions for each specific person, and the abnormal consumption frequency is used to form the fourth financial behavior feature.

[0021] The first financial behavior feature, the second financial behavior feature, the third financial behavior feature, and the fourth financial behavior feature constitute the financial behavior feature.

[0022] In one possible implementation, converting the traffic behavior data into traffic behavior features includes:

[0023] For the frequency data of the specific person taking public transportation in a certain functional area, the number of times the specific person takes public transportation at each starting transportation station is counted. Based on the number of times and the functional area to which each starting transportation station belongs, the probability of the specific person taking public transportation in each functional area is calculated. The probabilities of all functional areas are combined to form the first traffic behavior feature.

[0024] For the passenger flow data of the traffic stations, the passenger flow of each traffic station at different time periods is used to form a second traffic behavior feature;

[0025] For the functional area data of the traffic stations, the code of the functional area to which each traffic station belongs is used to form a third traffic behavior feature;

[0026] For the correlation data between the travel peak and the suspected gathering area, determine the probability that the peak business hours of the gathering place where the specific person is located match the peak travel hours of the surrounding transportation stations, and form the probability into a fourth traffic behavior feature.

[0027] The first traffic behavior feature, the second traffic behavior feature, the third traffic behavior feature, and the fourth traffic behavior feature are combined to form a traffic behavior feature.

[0028] In one possible implementation, training the RNN model based on the first training set includes:

[0029] The RNN model is used to process the financial behavior characteristics and traffic behavior characteristics in the first time period to obtain the predicted financial behavior characteristics and traffic behavior characteristics in the second time period.

[0030] The first loss value is calculated based on the predicted financial and traffic behavior characteristics of the second period and the actual financial and traffic behavior characteristics of the second period.

[0031] The model parameters of the RNN model are updated based on the first loss value.

[0032] In one possible implementation, training the PSO-ELM model based on the second training set includes:

[0033] The PSO-ELM model is used to process the financial and transportation behavior characteristics of the specific population to obtain a predicted value of the probability that the specific person will appear in a certain gathering place in a certain geographical area at a certain time.

[0034] The second loss value is calculated based on the predicted value and the annotation information;

[0035] The model parameters of the PSO-ELM model are updated based on the second loss value.

[0036] In one possible implementation, the method further includes:

[0037] Obtain historical financial and transportation behavior characteristics of the specific individuals to be predicted;

[0038] The RNN model is used to process the historical financial behavior characteristics and historical traffic behavior characteristics, and to predict the future financial behavior characteristics and future traffic behavior characteristics of the specific individuals.

[0039] The PSO-ELM model is used to process the historical financial behavior characteristics, the historical traffic behavior characteristics, the future financial behavior characteristics, and the future traffic behavior characteristics to obtain the probability that the specific person will appear in a certain gathering place in a certain geographical area in the future.

[0040] According to a second aspect of this application, a training apparatus for a model of a gathering place for specific individuals is provided, the apparatus comprising:

[0041] The acquisition module is used to acquire financial behavior characteristics, traffic behavior characteristics, and geographical region characteristics of a specific population.

[0042] The first generation module is used to generate a first training set based on the financial behavior characteristics and the traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and the financial behavior characteristics and traffic behavior characteristics in a second time period, wherein the second time period is later than the first time period.

[0043] The first training module is used to train a recurrent neural network (RNN) model based on the first training set.

[0044] The second generation module is used to generate a second training set based on the financial behavior characteristics, the traffic behavior characteristics, and the geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics, and annotation information of each specific person. The annotation information represents the probability that the specific person appears in a certain gathering place in a certain geographical region at a certain time.

[0045] The second training module is used to train the Particle Swarm Optimization (PSO)-Extreme Learning Machine (ELM) model based on the second training set.

[0046] According to a third aspect of this application, a computer-readable storage medium is provided, wherein at least one instruction is stored therein, the at least one instruction being loaded and executed by a processor to implement a training method for a gathering place model of specific persons as described above.

[0047] According to a fourth aspect of this application, a computer device is provided, the computer device including a training apparatus for a model of a gathering place of the aforementioned specific persons.

[0048] The beneficial effects of the technical solution provided in this application include at least the following:

[0049] The training samples in the first training set include the financial and traffic behavior characteristics of each specific individual in the first time period and the second time period, which is later than the first time period. After training the RNN model based on the first training set, the RNN model can predict future financial and traffic behavior characteristics based on the historical financial and traffic behavior characteristics of specific individuals, thereby expanding the behavioral characteristics of specific individuals, providing more analytical data for predicting the behavior of specific individuals, and improving the accuracy of prediction.

[0050] The PSO-ELM model utilizes the PSO algorithm to optimize the parameters in the ELM. Specifically, it treats the connection weights between the input and hidden layers in the ELM and the thresholds of the hidden layer neurons as particle positions in the PSO algorithm. Through the iterative search process of the PSO algorithm, an optimal set of parameters is found, thereby improving the performance of the ELM. This combination approach can overcome the performance instability problem caused by the random initialization of the ELM and improve the generalization ability and prediction accuracy of the PSO-ELM model. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart of a training method for a model of a gathering place of specific people provided in one embodiment of this application;

[0053] Figure 2 This is a flowchart of a training method for a model of a gathering place of specific people provided in one embodiment of this application;

[0054] Figure 3 This is a flowchart of a training method for a model of a gathering place of specific people provided in one embodiment of this application;

[0055] Figure 4 This is a structural block diagram of a training device for a specific personnel gathering place model provided in one embodiment of this application. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0057] like Figure 1 The diagram illustrates a method flowchart for training a model of a gathering place for specific individuals, according to an embodiment of this application. This method can be applied to a computer device. The method for training the model of a gathering place for specific individuals may include:

[0058] Step 101: Obtain the financial behavior characteristics, transportation behavior characteristics, and geographical region characteristics of a specific population.

[0059] A specific group is a group of specific people.

[0060] Financial behavior characteristics are features extracted based on the financial behaviors of specific individuals. The specific extraction methods are described in detail below.

[0061] Traffic behavior features are features extracted based on the traffic behaviors of specific individuals. The specific extraction methods are described in detail below.

[0062] Geographic region features are features extracted based on information about each geographic region. The specific extraction methods are described in detail below.

[0063] Step 102: Generate a first training set based on financial behavior characteristics and traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in the first time period and the financial behavior characteristics and traffic behavior characteristics in the second time period, which is later than the first time period.

[0064] For each specific individual, all financial and transportation behavior characteristics can be used to form a training sample, selecting financial and transportation behavior characteristics from a first time period and a second time period based on the time of occurrence of the behavior. For example, the first time period is January 1, 2024, and the second time period is February 5, 2024.

[0065] Step 103: Train the RNN model based on the first training set.

[0066] The RNN model is a type of recurrent neural network (RNN) model.

[0067] When training the RNN model using each training sample, the financial behavior features and traffic behavior features in the first time period are used as inputs to the RNN model, and the financial behavior features and traffic behavior features in the second time period are used as inputs to the loss function in the RNN model, so as to adjust the model parameters of the RNN model according to the output of the loss function.

[0068] Step 104: Generate a second training set based on financial behavior characteristics, traffic behavior characteristics, and geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics, and annotation information of each specific person. The annotation information indicates the probability that a specific person will appear in a certain gathering place in a certain geographical region at a certain time.

[0069] For example, if a specific person frequently visits a bar near a public transportation station late at night and spends a relatively large amount each time, the probability value in the labeling information can be set to a higher value. Conversely, if a specific person eats at a restaurant near a public transportation station at lunchtime and spends a relatively small amount each time, the probability value in the labeling information can be set to a lower value.

[0070] Step 105: Train the PSO-ELM model based on the second training set.

[0071] The PSO-ELM model is a model composed of the Particle Swarm Optimization (PSO) algorithm and the Extreme Learning Machine (ELM).

[0072] When training the PSO-ELM model using each training sample, financial behavior features, traffic behavior features, and geographical region features are used as inputs to the PSO-ELM model, and the labeled information is used as inputs to the loss function in the PSO-ELM model, so as to adjust the model parameters of the PSO-ELM model according to the output of the loss function.

[0073] In summary, the training method for the gathering place model of specific personnel provided in this application embodiment includes training samples in the first training set that include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and in a second time period, where the second time period is later than the first time period. After training the RNN model based on the first training set, the RNN model can predict future financial behavior characteristics and future traffic behavior characteristics based on the historical financial behavior characteristics and historical traffic behavior characteristics of specific personnel, thereby expanding the behavioral characteristics of specific personnel, providing more analytical data for predicting the behavior of specific personnel, and improving the accuracy of prediction.

[0074] The PSO-ELM model utilizes the PSO algorithm to optimize the parameters in the ELM. Specifically, it treats the connection weights between the input and hidden layers in the ELM and the thresholds of the hidden layer neurons as particle positions in the PSO algorithm. Through the iterative search process of the PSO algorithm, an optimal set of parameters is found, thereby improving the performance of the ELM. This combination approach can overcome the performance instability problem caused by the random initialization of the ELM and improve the generalization ability and prediction accuracy of the PSO-ELM model.

[0075] like Figure 2 The diagram illustrates a flowchart of a training method for a specific personnel gathering place model according to an embodiment of this application. This training method can be applied to a computer device. The training method for the specific personnel gathering place model may include:

[0076] Step 201: Obtain financial behavior data, traffic behavior data, and geographic area data for a specific population. Financial behavior data includes at least one of the following: consumption location data, consumption time data, consumption amount data, and abnormal consumption frequency data. Traffic behavior data includes data on the frequency of specific individuals taking public transportation in a certain functional area, passenger flow data entering and exiting transportation stations, data on the functional area to which the transportation station belongs, and correlation data between peak travel times and suspicious gathering places. Geographic area data includes at least one of the following: geographic data of each geographic area, population flow data, and historical crime data within the geographic area.

[0077] Functional zones refer to the division of a geographical area according to its function, such as commercial zones, residential zones, entertainment zones, industrial zones, etc.

[0078] Step 202: Convert financial behavior data into financial behavior features.

[0079] Specifically, converting financial behavior data into financial behavior characteristics can include:

[0080] (1) For consumption venue data, the code of each consumption venue category is used to form the first financial behavior feature.

[0081] The dimension of the first financial behavior characteristic is equal to the number of categories of consumption venues. For example, if consumption venues are divided into 3 categories, then the dimension of the first financial behavior characteristic is equal to 3, meaning that the first financial behavior characteristic contains 3 elements.

[0082] Assume that the categories of consumption venues include restaurants (denoted as the first element in the first financial behavior feature), entertainment venues (such as bars, KTVs, etc., denoted as the second element in the first financial behavior feature), and retail stores (denoted as the third element in the first financial behavior feature).

[0083] Suppose that the consumption venue is coded into 3 categories, and the first financial behavior feature of a specific person is represented in vector form [0,0,1], where the third bit is 1, indicating that the main consumption venue of the specific person belongs to the retail store category.

[0084] (2) For consumption time data, the coding of the time period to which each consumption time belongs is used to form the second financial behavior feature.

[0085] Consumption time data is used to record concentrated periods of consumption. The dimension of the second financial behavior characteristic is equal to the number of time periods. For example, if the time periods are divided into 3 categories, then the dimension of the second financial behavior characteristic is 3, meaning that the second financial behavior characteristic contains 3 elements.

[0086] Assume that the consumption time periods are divided into daytime (06:00-18:00, denoted as the first element in the second financial behavior feature), evening (18:00-24:00, denoted as the second element in the second financial behavior feature), and late night (00:00-06:00, denoted as the third element in the second financial behavior feature).

[0087] Suppose that the consumption time is divided into 3 time periods, and the second financial behavior feature of a specific person is represented by a vector [0,1,0], where the second bit is 1, indicating that the specific person's consumption is mainly concentrated in the evening.

[0088] (3) For consumption amount data, the code of the consumption level to which each consumption amount belongs is used to form the third financial behavior feature.

[0089] Consumption amount data is used to statistically analyze the frequency of consumption at different consumption levels (amount ranges). The dimension of the third financial behavior characteristic is equal to the number of consumption levels. For example, if there are 3 consumption levels, then the dimension of the third financial behavior characteristic is 3, meaning that the third financial behavior characteristic contains 3 elements.

[0090] Assume that consumption levels include small consumption (0-100 yuan, denoted as the first element in the third financial behavior characteristic), medium consumption (100-500 yuan, denoted as the second element in the third financial behavior characteristic), and large consumption (above 500 yuan, denoted as the third element in the third financial behavior characteristic).

[0091] Suppose that consumption levels are divided into 3 levels, and the third financial behavior characteristics of a specific person are represented by a vector [0.3, 0.4, 0.3], which means that the frequency of small, medium and large consumption of a specific person is 0.3, 0.4 and 0.3 respectively.

[0092] (4) For abnormal consumption frequency data, the abnormal consumption frequency is calculated based on the number of abnormal consumptions and the total number of consumptions for each specific person, and the abnormal consumption frequency is used to form the fourth financial behavior feature.

[0093] Abnormal spending frequency data is used to identify the proportion of abnormal spending that does not conform to the average spending habits of an individual or geographic region to the total number of spending transactions.

[0094] The fourth financial behavior characteristic can be represented by a numerical value, such as 0.2, which indicates that the frequency of abnormal consumption is 20%.

[0095] (5) The first financial behavior feature, the second financial behavior feature, the third financial behavior feature and the fourth financial behavior feature are combined to form financial behavior features.

[0096] The computer device can set the splicing order of the first financial behavior feature, the second financial behavior feature, the third financial behavior feature, and the fourth financial behavior feature, and splice the first financial behavior feature, the second financial behavior feature, the third financial behavior feature, and the fourth financial behavior feature into a financial behavior feature according to the splicing order.

[0097] Step 203: Convert traffic behavior data into traffic behavior features.

[0098] Specifically, converting traffic behavior data into traffic behavior features can include:

[0099] (1) For the frequency data of a specific person taking public transportation in a certain functional area, count the number of times the specific person takes public transportation at each starting traffic station, calculate the probability of the specific person taking public transportation in each functional area based on the number of times and the functional area to which each starting traffic station belongs, and form the first traffic behavior feature by combining the probabilities of all functional areas.

[0100] Suppose functional areas are divided into four categories: commercial areas (denoted as the first element of the first traffic behavior characteristic), residential areas (denoted as the second element of the first traffic behavior characteristic), entertainment areas (denoted as the third element of the first traffic behavior characteristic), and industrial areas (denoted as the fourth element of the first traffic behavior characteristic). Computer equipment can classify the starting stations of public transportation such as subways and buses according to the type of functional area. Assuming there are 10 different starting stations, the frequency of occurrence of the functional area to which each starting station belongs can be counted.

[0101] Suppose that the first traffic behavior characteristic of a specific person is represented by a vector [0.2, 0.3, 0.4, 0.1]. This means that in the statistical ride records, the frequency of the starting traffic station being a commercial area is 0.2, the probability of being a residential area is 0.3, the probability of being an entertainment area is 0.4, and the probability of being an industrial area is 0.1.

[0102] (2) For the passenger flow data of entering and exiting transportation stations, the passenger flow of each transportation station at different time periods is used to form a second traffic behavior feature.

[0103] Computer equipment tracks the flow of people entering and exiting each transportation station (bus stop, subway station) at different times.

[0104] (3) For the functional area data of the traffic station, the code of the functional area to which each traffic station belongs is used to form the third traffic behavior feature.

[0105] Suppose functional areas are divided into four categories: commercial areas (denoted as the first element of the third traffic behavior characteristic), residential areas (denoted as the second element of the third traffic behavior characteristic), entertainment areas (denoted as the third element of the third traffic behavior characteristic), and industrial areas (denoted as the fourth element of the third traffic behavior characteristic). Computer equipment can classify public transportation stations such as subways and buses according to the type of functional area.

[0106] (4) For the correlation data between travel peaks and suspicious gathering areas, determine the probability that the peak business hours of the gathering place where a specific person is located match the peak travel hours of the surrounding transportation stations, and combine the probability into the fourth traffic behavior feature.

[0107] Record whether peak traffic periods match suspicious geographic areas where specific people may gather (determined based on historical data or other clues).

[0108] (5) The first traffic behavior feature, the second traffic behavior feature, the third traffic behavior feature and the fourth traffic behavior feature are combined to form a traffic behavior feature.

[0109] The computer equipment can set the splicing order of the first traffic behavior feature, the second traffic behavior feature, the third traffic behavior feature, and the fourth traffic behavior feature, and splice the first traffic behavior feature, the second traffic behavior feature, the third traffic behavior feature, and the fourth traffic behavior feature into a traffic behavior feature according to the splicing order.

[0110] Step 204: Convert the geographic region data into geographic region features.

[0111] (1) When geographic region data includes geographic information:

[0112] The first geographic region feature is generated based on the coordinates of each block within the geographic area, such as (10.2, 20.5). Because the gathering places of specific people may be related to geographical location, some remote areas or areas with relatively weak security management may be more likely to become specific places. These coordinates can help identify different geographic regions.

[0113] A second geographic region feature is generated based on the distance of the geographic region from entertainment venues (bars, nightclubs, etc.), expressed in meters, such as 500 meters, 1000 meters, etc. The area surrounding entertainment venues may have a higher probability of specific groups of people congregating due to factors such as complex population flow.

[0114] A third geographic region feature is generated based on the distance of the geographic region from public transportation stations (bus stops, subway stations), such as 300 meters, 700 meters, etc. Locations with convenient transportation may facilitate the entry and exit of specific individuals, thus increasing their likelihood of becoming gathering places.

[0115] (2) When geographic region data includes population movement data:

[0116] A fourth geographic region feature is generated based on pedestrian traffic in various geographic areas at different times (e.g., daytime, nighttime). Pedestrian traffic can be the number of people passing through per hour, such as 1000 people / hour, 500 people / hour, etc. Areas with low pedestrian traffic and poor regulation may be more prone to specific groups of people gathering.

[0117] The fifth geographic region characteristic is generated based on the proportion of migrant population within a geographic region. The proportion of migrant population can be expressed as a percentage, such as 20%, 30%, etc. Geographic regions with a large migrant population may have their probability of specific individuals clustering affected due to factors such as complex population structure.

[0118] (3) When geographic area data includes historical crime data:

[0119] A sixth geographic region characteristic is generated based on the number of specific behavioral crimes committed within the geographic area over the past year, such as 0 cases, 3 cases, etc. Geographic areas with a historically high incidence of specific behavioral crimes are likely to be gathering places for specific individuals.

[0120] The seventh geographic region feature is generated based on the number of other criminal cases (such as theft and robbery) within the geographic region. Because criminal acts are often related, geographic regions with high crime rates may also have a relatively high concentration of specific individuals, with the number of such cases potentially being 5, 10, or similar.

[0121] The computer equipment combines the features of the first, second, third, fourth, fifth, sixth, and seventh geographical regions into a single geographical feature.

[0122] Step 205: Generate a first training set based on financial behavior characteristics and traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in the first time period and the financial behavior characteristics and traffic behavior characteristics in the second time period, which is later than the first time period.

[0123] The generation of the first training set is described in step 102, and will not be repeated here.

[0124] Step 206: Train the RNN model based on the first training set.

[0125] When training the RNN model using each training sample, the financial behavior features and traffic behavior features in the first time period are used as inputs to the RNN model, and the financial behavior features and traffic behavior features in the second time period are used as inputs to the loss function in the RNN model, so as to adjust the model parameters of the RNN model according to the output of the loss function.

[0126] Specifically, training an RNN model based on the first training set may include: processing the financial behavior features and traffic behavior features in the first time period using the RNN model to obtain the predicted financial behavior features and traffic behavior features in the second time period; calculating a first loss value based on the predicted financial behavior features and traffic behavior features in the second time period and the actual financial behavior features and traffic behavior features in the second time period; and updating the model parameters of the RNN model based on the first loss value.

[0127] Step 207: Generate a second training set based on financial behavior characteristics, traffic behavior characteristics, and geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics, and annotation information of each specific person. The annotation information represents the probability that a specific person will appear in a certain gathering place in a certain geographical region at a certain time.

[0128] The generation of the second training set is described in step 104, and will not be repeated here.

[0129] Step 208: Train the PSO-ELM model based on the second training set.

[0130] When training the PSO-ELM model using each training sample, financial behavior features, traffic behavior features, and geographical region features are used as inputs to the PSO-ELM model, and the labeled information is used as inputs to the loss function in the PSO-ELM model, so as to adjust the model parameters of the PSO-ELM model according to the output of the loss function.

[0131] Specifically, training the PSO-ELM model based on the second training set may include: processing the financial and traffic behavior characteristics of a specific group of people using the PSO-ELM model to obtain a predicted value of the probability that a specific person will appear in a certain gathering place in a certain geographical area at a certain time; calculating a second loss value based on the predicted value and the annotation information; and updating the model parameters of the PSO-ELM model based on the second loss value.

[0132] In summary, the training method for the gathering place model of specific personnel provided in this application embodiment includes training samples in the first training set that include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and in a second time period, where the second time period is later than the first time period. After training the RNN model based on the first training set, the RNN model can predict future financial behavior characteristics and future traffic behavior characteristics based on the historical financial behavior characteristics and historical traffic behavior characteristics of specific personnel, thereby expanding the behavioral characteristics of specific personnel, providing more analytical data for predicting the behavior of specific personnel, and improving the accuracy of prediction.

[0133] The PSO-ELM model utilizes the PSO algorithm to optimize the parameters in the ELM. Specifically, it treats the connection weights between the input and hidden layers in the ELM and the thresholds of the hidden layer neurons as particle positions in the PSO algorithm. Through the iterative search process of the PSO algorithm, an optimal set of parameters is found, thereby improving the performance of the ELM. This combination approach can overcome the performance instability problem caused by the random initialization of the ELM and improve the generalization ability and prediction accuracy of the PSO-ELM model.

[0134] like Figure 3 The diagram illustrates a flowchart of a method for using a specific personnel gathering place model according to an embodiment of this application. This method can be applied to a computer device. The method for using the specific personnel gathering place model may include:

[0135] Step 301: Obtain the historical financial behavior characteristics and historical traffic behavior characteristics of the specific person to be predicted.

[0136] Computer equipment can acquire the historical financial behavior and historical traffic behavior of specific individuals, extract historical financial behavior features from historical financial behavior, and extract historical traffic behavior data from historical traffic behavior. For details of the extraction methods, please refer to the descriptions in steps 202 and 203, which will not be repeated here.

[0137] Step 302: Use an RNN model to process historical financial behavior characteristics and historical traffic behavior characteristics, and predict the future financial behavior characteristics and future traffic behavior characteristics of specific individuals.

[0138] The computer device inputs historical financial behavior characteristics and historical traffic behavior characteristics into the RNN model, and determines the future financial behavior characteristics and future traffic behavior characteristics from the output of the RNN model.

[0139] Step 303: Use the PSO-ELM model to process historical financial behavior characteristics, historical traffic behavior characteristics, future financial behavior characteristics, and future traffic behavior characteristics to obtain the probability that a specific person will appear in a certain gathering place in a certain geographical area in the future.

[0140] The computer device inputs historical financial behavior characteristics, historical traffic behavior characteristics, future financial behavior characteristics, and future traffic behavior characteristics into the PSO-ELM model, and determines the output of the PSO-ELM model as the probability that a specific person will appear in a certain gathering place in a certain geographical area in the future.

[0141] like Figure 4 The diagram illustrates a structural block diagram of a training apparatus for a specific personnel gathering place model according to an embodiment of this application. This training apparatus for the specific personnel gathering place model can be applied to a computer device. The training apparatus for the specific personnel gathering place model may include:

[0142] The acquisition module 410 is used to acquire the financial behavior characteristics, traffic behavior characteristics, and geographical region characteristics of a specific population.

[0143] The first generation module 420 is used to generate a first training set based on financial behavior characteristics and traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and the financial behavior characteristics and traffic behavior characteristics in a second time period, which is later than the first time period.

[0144] The first training module 430 is used to train a recurrent neural network (RNN) model based on the first training set.

[0145] The second generation module 440 is used to generate a second training set based on financial behavior characteristics, traffic behavior characteristics and geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics and annotation information of each specific person. The annotation information indicates the probability that a specific person will appear in a certain gathering place in a certain geographical region at a certain time.

[0146] The second training module 450 is used to train the PSO-ELM model based on the second training set.

[0147] In an optional embodiment, the acquisition module 410 is further configured to:

[0148] Acquire financial behavior data, traffic behavior data, and geographic area data for specific groups. Financial behavior data includes at least one of the following: consumption location data, consumption time data, consumption amount data, and abnormal consumption frequency data. Traffic behavior data includes data on the frequency of specific individuals taking public transportation in a certain functional area, passenger flow data entering and exiting transportation stations, data on the functional area to which transportation stations belong, and correlation data between peak travel times and suspicious gathering places. Geographic area data includes at least one of the following: geographic data of each geographic area, population flow data, and historical crime data within the geographic area.

[0149] Transform financial behavior data into financial behavior characteristics;

[0150] Transform traffic behavior data into traffic behavior characteristics;

[0151] Transform geographic region data into geographic region features.

[0152] In an optional embodiment, the acquisition module 410 is further configured to:

[0153] For consumption venue data, the code of each consumption venue's category is used to form the first financial behavior feature;

[0154] For consumption time data, the coding of the time period to which each consumption time belongs is used to form the second financial behavior feature;

[0155] For consumption amount data, the code of the consumption level to which each consumption amount belongs is used to form the third financial behavior feature;

[0156] For abnormal consumption frequency data, the abnormal consumption frequency is calculated based on the number of abnormal consumptions and the total number of consumptions for each specific person, and the abnormal consumption frequency is used to form the fourth financial behavior feature.

[0157] The first financial behavior characteristic, the second financial behavior characteristic, the third financial behavior characteristic, and the fourth financial behavior characteristic are combined to form the financial behavior characteristics.

[0158] In an optional embodiment, the acquisition module 410 is further configured to:

[0159] For the frequency data of a specific person taking public transportation in a certain functional area, the number of times the specific person takes public transportation at each starting transportation station is counted. Based on the number of times and the functional area to which each starting transportation station belongs, the probability of the specific person taking public transportation in each functional area is calculated. The probabilities of all functional areas are combined to form the first traffic behavior feature.

[0160] For the passenger flow data of entering and exiting transportation stations, the passenger flow of each transportation station at different time periods is used to form a second traffic behavior feature;

[0161] For the functional area data of each transportation station, the code of the functional area to which each transportation station belongs is used to form the third traffic behavior feature;

[0162] For the correlation data between peak travel times and suspicious gathering areas, determine the probability that the peak business hours of a gathering place where a specific person is located match the peak travel times of surrounding transportation stations, and combine the probabilities into a fourth traffic behavior feature.

[0163] The first traffic behavior feature, the second traffic behavior feature, the third traffic behavior feature, and the fourth traffic behavior feature are combined to form the traffic behavior feature.

[0164] In an optional embodiment, the first training module 430 is further configured to:

[0165] The financial and traffic behavior characteristics in the first time period are processed using an RNN model to obtain the predicted financial and traffic behavior characteristics in the second time period.

[0166] The first loss value is calculated based on the predicted financial and traffic behavior characteristics of the second period and the actual financial and traffic behavior characteristics of the second period.

[0167] Update the model parameters of the RNN model based on the first loss value.

[0168] In an optional embodiment, the second training module 450 is further configured to:

[0169] The PSO-ELM model is used to process the financial and transportation behavior characteristics of a specific group of people to obtain the predicted probability of a specific person appearing at a certain gathering place in a certain geographical area at a certain time.

[0170] Calculate the second loss value based on the predicted value and the labeled information;

[0171] The model parameters of the PSO-ELM model are updated based on the second loss value.

[0172] In an optional embodiment, the acquisition module 410 is further configured to acquire historical financial behavior characteristics and historical traffic behavior characteristics of the specific person to be predicted.

[0173] The prediction module is used to process historical financial behavior characteristics and historical traffic behavior characteristics using an RNN model, and to predict the future financial behavior characteristics and future traffic behavior characteristics of specific individuals.

[0174] The prediction module is also used to process historical financial behavior characteristics, historical traffic behavior characteristics, future financial behavior characteristics, and future traffic behavior characteristics using the PSO-ELM model to obtain the probability that a specific person will appear in a certain gathering place in a certain geographical area in the future.

[0175] In summary, the training device for the gathering place model of specific personnel provided in this application embodiment includes training samples in the first training set that include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and in a second time period, where the second time period is later than the first time period. After training the RNN model based on the first training set, the RNN model can predict future financial behavior characteristics and future traffic behavior characteristics based on the historical financial behavior characteristics and historical traffic behavior characteristics of specific personnel, thereby expanding the behavioral characteristics of specific personnel, providing more analytical data for predicting the behavior of specific personnel, and improving the accuracy of prediction.

[0176] The PSO-ELM model utilizes the PSO algorithm to optimize the parameters in the ELM. Specifically, it treats the connection weights between the input and hidden layers in the ELM and the thresholds of the hidden layer neurons as particle positions in the PSO algorithm. Through the iterative search process of the PSO algorithm, an optimal set of parameters is found, thereby improving the performance of the ELM. This combination approach can overcome the performance instability problem caused by the random initialization of the ELM and improve the generalization ability and prediction accuracy of the PSO-ELM model.

[0177] One embodiment of this application provides a computer-readable storage medium storing at least one instruction, which is loaded and executed by a processor to implement the training method for a gathering place model of specific people as described above.

[0178] One embodiment of this application provides a computer device that includes a training apparatus for a model of a gathering place of any specific persons.

[0179] It should be noted that the training device for the gathering place model of specific personnel provided in the above embodiments is only illustrated by the division of the above functional modules when training the gathering place model of specific personnel. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the training device for the gathering place model of specific personnel can be divided into different functional modules to complete all or part of the functions described above. In addition, the training device for the gathering place model of specific personnel provided in the above embodiments and the training method embodiment for the gathering place model of specific personnel belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0180] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0181] The above description is not intended to limit the embodiments of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of the embodiments of this application.

Claims

1. A training method for a model of a gathering place for specific individuals, characterized in that, The method includes: To obtain the financial behavior characteristics, transportation behavior characteristics, and geographical region characteristics of a specific population; A first training set is generated based on the financial behavior characteristics and the traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and in a second time period, wherein the second time period is later than the first time period. Train a recurrent neural network (RNN) model based on the first training set; A second training set is generated based on the financial behavior characteristics, the traffic behavior characteristics, and the geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics, and annotation information of each specific person. The annotation information represents the probability that the specific person appears in a certain gathering place in a certain geographical region at a certain time. Train the Particle Swarm Optimization (PSO)-Extreme Learning Machine (ELM) model using the second training set; The method further includes: Obtain historical financial and transportation behavior characteristics of the specific individuals to be predicted; The RNN model is used to process the historical financial behavior characteristics and historical traffic behavior characteristics, and to predict the future financial behavior characteristics and future traffic behavior characteristics of the specific individuals. The PSO-ELM model is used to process the historical financial behavior characteristics, the historical traffic behavior characteristics, the future financial behavior characteristics, and the future traffic behavior characteristics to obtain the probability that the specific person will appear in a certain gathering place in a certain geographical area in the future.

2. The training method for a specific personnel gathering place model according to claim 1, characterized in that, The acquisition of financial behavior characteristics, transportation behavior characteristics, and geographical region characteristics of specific population groups includes: The system acquires financial behavior data, traffic behavior data, and geographic area data for a specific population group. The financial behavior data includes at least one of the following: consumption location data, consumption time data, consumption amount data, and abnormal consumption frequency data. The traffic behavior data includes data on the frequency of specific individuals using public transportation in a certain functional area, passenger flow data entering and exiting transportation stations, data on the functional area to which the transportation stations belong, and correlation data between peak travel times and suspicious gathering places. The geographic area data includes at least one of the following: geographic data of each geographic area, population flow data, and historical crime data within the geographic area. Convert the financial behavior data into financial behavior characteristics; The traffic behavior data is converted into traffic behavior features; The geographic region data is converted into geographic region features.

3. The training method for a specific personnel gathering place model according to claim 2, characterized in that, The process of converting the financial behavior data into financial behavior features includes: For the aforementioned consumption venue data, the code of the category to which each consumption venue belongs is used to form the first financial behavior feature; For the consumption time data, the encoding of the time period to which each consumption time belongs is used to form a second financial behavior feature; For the aforementioned consumption amount data, the codes of the consumption level to which each consumption amount belongs are used to form a third financial behavior feature; For the abnormal consumption frequency data, the abnormal consumption frequency is calculated based on the number of abnormal consumptions and the total number of consumptions for each specific person, and the abnormal consumption frequency is used to form the fourth financial behavior feature. The first financial behavior feature, the second financial behavior feature, the third financial behavior feature, and the fourth financial behavior feature constitute the financial behavior feature.

4. The training method for a specific personnel gathering place model according to claim 2, characterized in that, The process of converting the traffic behavior data into traffic behavior features includes: For the frequency data of the specific person taking public transportation in a certain functional area, the number of times the specific person takes public transportation at each starting transportation station is counted. Based on the number of times and the functional area to which each starting transportation station belongs, the probability of the specific person taking public transportation in each functional area is calculated. The probabilities of all functional areas are combined to form the first traffic behavior feature. For the passenger flow data of the traffic stations, the passenger flow of each traffic station at different time periods is used to form a second traffic behavior feature; For the functional area data of the traffic stations, the code of the functional area to which each traffic station belongs is used to form a third traffic behavior feature; For the correlation data between the travel peak and the suspected gathering area, determine the probability that the peak business hours of the gathering place where the specific person is located match the peak travel hours of the surrounding transportation stations, and form the probability into a fourth traffic behavior feature. The first traffic behavior feature, the second traffic behavior feature, the third traffic behavior feature, and the fourth traffic behavior feature are combined to form a traffic behavior feature.

5. The training method for a specific personnel gathering place model according to claim 1, characterized in that, The step of training the RNN model based on the first training set includes: The RNN model is used to process the financial behavior characteristics and traffic behavior characteristics in the first time period to obtain the predicted financial behavior characteristics and traffic behavior characteristics in the second time period. The first loss value is calculated based on the predicted financial and traffic behavior characteristics of the second period and the actual financial and traffic behavior characteristics of the second period. The model parameters of the RNN model are updated based on the first loss value.

6. The training method for a specific personnel gathering place model according to claim 1, characterized in that, The step of training the PSO-ELM model based on the second training set includes: The PSO-ELM model is used to process the financial and transportation behavior characteristics of the specific population to obtain a predicted value of the probability that the specific person will appear in a certain gathering place in a certain geographical area at a certain time. Calculate the second loss value based on the predicted value and the annotation information; The model parameters of the PSO-ELM model are updated based on the second loss value.

7. A training device for a model of a gathering place for specific personnel, characterized in that, The device includes: The acquisition module is used to acquire financial behavior characteristics, traffic behavior characteristics, and geographical region characteristics of specific groups of people. The first generation module is used to generate a first training set based on the financial behavior characteristics and the traffic behavior characteristics. The training samples in the first training set include the financial behavior characteristics and traffic behavior characteristics of each specific person in a first time period and the financial behavior characteristics and traffic behavior characteristics in a second time period, wherein the second time period is later than the first time period. The first training module is used to train a recurrent neural network (RNN) model based on the first training set. The second generation module is used to generate a second training set based on the financial behavior characteristics, the traffic behavior characteristics, and the geographical region characteristics. The training samples in the second training set include the financial behavior characteristics, traffic behavior characteristics, geographical region characteristics, and annotation information of each specific person. The annotation information represents the probability that the specific person appears in a certain gathering place in a certain geographical region at a certain time. The second training module is used to train the particle swarm optimization algorithm PSO-extreme learning machine ELM model based on the second training set. The acquisition module is also used to acquire the historical financial behavior characteristics and historical traffic behavior characteristics of the specific person to be predicted; The device also includes a prediction module, which uses the RNN model to process the historical financial behavior characteristics and historical traffic behavior characteristics, and predicts the future financial behavior characteristics and future traffic behavior characteristics of the specific person. The prediction module is further used to process the historical financial behavior characteristics, the historical traffic behavior characteristics, the future financial behavior characteristics, and the future traffic behavior characteristics using the PSO-ELM model to obtain the probability that the specific person will appear in a certain gathering place in a certain geographical area in the future.

8. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, which is loaded and executed by a processor to implement the training method for a gathering place model of specific personnel as described in any one of claims 1 to 6.

9. A computer device, characterized in that, The computer equipment includes: a training device for a model of a gathering place of specific personnel as described in claim 7.

Images