Heuristic data collection method, device, computer equipment and application thereof based on evolutionary game

Through the heuristic data acquisition method based on evolutionary game, the use of profit incentives and dynamic strategy adjustments, the problem of inefficiency of traditional data acquisition methods is solved, and an efficient and resource-saving data acquisition process is realized, and data quality and coverage are improved.

CN112016047BActive Publication Date: 2024-11-08ZHEJIANG UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010727542.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-07-24
Publication Date
2024-11-08
Estimated Expiration
2040-07-24

AI Technical Summary

Technical Problem

Traditional data acquisition methods are inefficient on smartphone devices, have serious resource waste, and it is difficult to control the data acquisition process in complex environments.

Method used

The heuristic data acquisition method based on evolutionary game is adopted, and the revenue incentive mechanism and dynamic strategy adjustments are adjusted to encourage users to generate effective data acquisition, and the acquisition strategy is adjusted through user status information and average income calculation to achieve Nash balance.

Benefits of technology

It improves data acquisition efficiency, saves resources, controls the data acquisition process, obtains better data sets, and improves the coverage of data application scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112016047B_ABST
    Figure CN112016047B_ABST
Patent Text Reader

Abstract

The present invention provides a heuristic data collection method, device, computer equipment and application based on evolutionary game, which are applied to a heuristic data collection system including participating users and servers. In the data collection method, the server receives the sensor data collected by the participating users according to the sensor process of data collection in the access request; the benefits calculated for the received sensor data, and the strategy adjustment plan formulated for the participating users according to the status information of the participating users and the average benefits of all participating users of the heuristic data collection system, and sent to the participating users, and the strategy adjustment plan is used as a reference for the participating users to adjust the sensor process of data collection. The present invention motivates users to participate in the provision of device collection data through a benefit mechanism, and realizes heuristic data collection through a participatory game mechanism, thereby improving the collection efficiency of horizontal data and the coverage of data application scenarios.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data collection, and in particular to a heuristic data collection method, device, computer equipment based on evolutionary game, and application thereof in mobile device user authentication. Background Art

[0002] With the development of modern science and technology, artificial intelligence technology has penetrated into all walks of life and various technical fields. Machine learning and deep learning technology methods are also constantly innovating. Its essence is to generate models through training with a large amount of data and make decisions through models. In the whole process, data is the most important foundation for training and the root of everything.

[0003] The traditional data collection method for collecting sensor data from smartphone devices is to invite a small number of participating users and require them to collect new data for each action. This method is very wasteful of manpower and material resources, resulting in low collection efficiency. In addition, in a complex actual environment, when users are required to collect certain specific data, they cannot perform specific operations or fix the position of their mobile phones for a long time, so data collection takes a long time and the collection process cannot be controlled.

[0004] Game theory is a mathematical theory and method for studying the interaction of formalized incentive structures and phenomena of a combative or competitive nature. Using incentives and dynamic strategy adjustments in games to collect sensor data from users' smartphone devices is a solution. Summary of the invention

[0005] In order to overcome the problems of resource waste and uncontrollable data collection process, the present invention provides a data collection method based on evolutionary game mechanism to continuously provide the best strategy. By generating revenue incentives, users are encouraged to generate effective data for collection, and the collection strategy is continuously adjusted through the user's usage data to achieve Nash equilibrium.

[0006] The present invention discloses a heuristic data collection method based on evolutionary game, which is applied to a heuristic data collection system including participating users and a server, and performs the following operations on the server side:

[0007] Receiving an access request sent by a participating user to a server, wherein the access request includes a data collection sensing process selected from a preset data collection sensing process set;

[0008] Sending a response to the received access request to the participating user and allowing the participating user to access;

[0009] Receiving sensor data collected by the participating user according to the sensor process of data collection in the access request;

[0010] A strategy adjustment plan is formulated for the participating user based on the benefits calculated for the received sensor data and the status information of the participating user and the average benefits of all participating users of the heuristic data collection system, and sent to the participating user, wherein the strategy adjustment plan is used as a reference for the participating user to adjust the sensing process of data collection;

[0011] The status information of the participating users includes the participating users' own benefits.

[0012] Preferably, the benefit calculated for the received sensor data is performed according to Φ:

[0013]

[0014] Among them, the set of sensing processes P, the profit calculation method, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed investment budget of sensing process p, N represents the number of users who choose sensing process p at time t, p∈P.

[0015] Preferably, the strategy adjustment scheme formulated for the participating users according to the status information of the participating users and the average benefits of all participating users of the heuristic data acquisition system includes:

[0016] For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval;

[0017] The difference δ is calculated according to the following formula * :

[0018]

[0019] Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the revenue of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P];

[0020] According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

[0021] The present invention also provides a heuristic data collection method based on evolutionary game, which is applied to a heuristic data collection system including participating users and a server, and performs the following operations at the participating user end:

[0022] Sending an access request to a server, wherein the access request includes a data collection and sensing process selected from a preset data collection and sensing process set;

[0023] receiving a response from the server to the access request, and accessing the server;

[0024] Collect sensor data according to the data collection and sensing process in the access request and send it to the server;

[0025] The receiving server calculates the benefits of the received sensing data and formulates a strategy adjustment plan for the participating users according to the average benefits of all participating users of the heuristic data collection system, wherein the strategy adjustment plan is used as a reference for the participating users to adjust the data collection sensing process.

[0026] The present invention also provides a heuristic data collection device based on evolutionary game, comprising:

[0027] The first unit is used to receive an access request sent by a participating user to a server, wherein the access request includes a data collection sensing process selected from a preset data collection sensing process set;

[0028] The second unit is used to send a response to the received access request to the participating user and allow the participating user to access;

[0029] A third unit is used to receive sensor data collected by the participating user according to the sensor process of data collection in the access request;

[0030] The fourth unit is used for calculating the benefits of the received sensor data, and formulating a strategy adjustment plan for the participating user according to the status information of the participating user and the average benefits of all participating users of the heuristic data collection system, and sending the strategy adjustment plan to the participating user, wherein the strategy adjustment plan is used as a reference for the participating user to adjust the sensing process of data collection;

[0031] The status information of the participating users includes the participating users' own benefits.

[0032] Preferably, the fourth unit calculates the benefit of the received sensor data according to Φ, and the benefit of the received sensor data is calculated according to Φ:

[0033]

[0034] Among them, the set of sensing processes P, the profit calculation method, θ(t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed investment budget of sensing process p, N represents the number of users who choose sensing process p at time t, p∈P.

[0035] Preferably, the fourth unit performs the following operations when formulating a strategy adjustment plan for a participating user according to the status information of the participating user and the average benefit of all participating users of the heuristic data acquisition system:

[0036] For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval;

[0037] The difference δ is calculated according to the following formula * :

[0038]

[0039] Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the revenue of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P];

[0040] According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

[0041] The present invention also provides a heuristic data collection device based on evolutionary game, comprising:

[0042] A fifth unit is used to send an access request to the server, wherein the access request includes a data collection and sensing process selected from a preset data collection and sensing process set;

[0043] A sixth unit is used to receive a response from the server to the access request and access the server;

[0044] A seventh unit is used to collect sensor data according to the data collection and sensing process in the access request and send it to the server;

[0045] The receiving server calculates the benefits of the received sensing data and formulates a strategy adjustment plan for the participating users according to the average benefits of all participating users of the heuristic data collection system, wherein the strategy adjustment plan is used as a reference for the participating users to adjust the data collection sensing process.

[0046] The present invention also provides an electronic device, comprising:

[0047] one or more processors;

[0048] A storage device for storing one or more programs;

[0049] When the one or more programs are executed by the one or more processors, the one or more processors implement the above method.

[0050] The present invention also provides a computer-readable medium on which a computer program is stored, and the program implements the above method when executed by a processor.

[0051] The advantages of the present invention are: (1) users are encouraged to participate in the provision of device data collection through a profit mechanism. (2) heuristic data collection is achieved through a participatory game mechanism, the efficiency of horizontal data collection is improved, and the coverage of data application scenarios is improved. (3) resource consumption in the data collection process is saved, and the data collection process is controlled to obtain a better data set. (4) the application of game theory can efficiently and stably collect the required data. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a framework diagram of the mobile device authentication system for end-cloud collaboration;

[0053] Figure 2 It is the optimized LSTM neural network structure diagram;

[0054] Figure 3 It is the enhanced SGD algorithm flow chart;

[0055] Figure 4 This is the interaction diagram between the user and the server of the data acquisition system.

[0056] Figure 5 The following is a flowchart of server-side dynamic policy adjustment. DETAILED DESCRIPTION

[0057] The present invention is further described below in conjunction with embodiments. The description of the following embodiments is only used to help understand the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principles of the present invention, several improvements and modifications may be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

[0058] In one embodiment of the present invention, a mobile device user authentication method based on optimized LSTM comprises the following steps:

[0059] Acquire a training sample set including a plurality of training samples, wherein the training samples include feature data of a mobile device for device user authentication and a label corresponding to the feature data;

[0060] Using the training sample set to train a pre-set LSTM model to obtain a certification model;

[0061] The characteristic data of the user of the mobile device to be authenticated is obtained as the input of the authentication model, and the output of the authentication model under the input is used as the authentication result of the device to be authenticated.

[0062] There is no special requirement for the size of the training sample set. When authenticating a user of a mobile device, the feature data includes data such as the acceleration sensor, gravity sensor, and gyroscope of the mobile device that can identify user features, including but not limited to the above data, or any combination of one or more of the above data.

[0063] Training samples can be obtained using existing means. In order to improve efficiency and ensure the effectiveness of data distribution, this application also provides a data collection method based on game theory, which is described below and will not be expanded here.

[0064] The method of using the training sample set to train a preset LSTM model to obtain a certification model includes: using an enhanced SGD algorithm to learn network parameters of the LSTM neural network based on the training sample set.

[0065] Learning the network parameters of the LSTM neural network is mainly divided into two steps, as follows: using the SGD algorithm to learn the LSTM neural network using the training sample set to obtain the gradient value of the model;

[0066] A gradient descent operation is performed based on the gradient value, and the optimal network parameters of the LSTM neural network are obtained by iterative calculation, so that the LSTM neural network corresponding to the optimal network parameters is used as the authentication model.

[0067] Specifically, learning the network parameters of the LSTM neural network using the enhanced SGD algorithm based on the training sample set includes:

[0068] Set the initial network parameters and the maximum number of iterations;

[0069] Based on the enhanced SGD algorithm to learn the network parameters of the LSTM neural network, each iteration performs the following operations:

[0070] Use random sampling probability n / N to randomly select samples from the training sample set to enter the group, where n is the number of randomly selected training samples and N is the number of samples in the training sample set;

[0071] Calculating the gradient value of the neural network using the sample entry group and a preset loss function;

[0072] Clipping the gradient value to obtain a clipped gradient value;

[0073] Add Gaussian noise as a robust factor to the clipped gradient value to get the robust gradient value

[0074] The gradient descent operation is performed according to the following formula:

[0075]

[0076] Where t is the current iteration number, t = 0, 1, 2, ... T, T is the maximum iteration number, η t is the learning rate for the preset t-th iteration, t=0, 1, 2, ... T, θ t The network parameters of the tth iteration, θ0 is the preset initial network parameters, is the robust gradient value calculated for the tth iteration.

[0077] By enhancing the SDG algorithm through new pruning and adding robust factors, the impact of the accuracy of the labels of each feature data in the training sample set on the final certification model readiness rate can be effectively reduced.

[0078] Use the sample to enter group G according to the following formula t And the preset loss function calculates the gradient value g t (x i ):

[0079]

[0080] in, is the preset loss function:

[0081]

[0082] Among them, x i is the i-th training sample in the sample entry group, θ represents the network parameters; y i Represents the i-th training sample x i The corresponding labels are 0 for other people’s data and 1 for personal data; f(x i , θ) indicates that the input is the training sample x i The classification value calculated when the network parameter is θ has a value range of 0 to 1.

[0083] The gradient value is clipped by standard clipping using the L1 norm to obtain the clipped gradient value:

[0084]

[0085] In the formula, g t (x i ) represents the gradient value, Represents the clipping gradient value, which is the clipping threshold and is used to control the gradient norm boundary.

[0086] A mobile device user authentication device based on optimized LSTM of the present invention comprises:

[0087] A first module is used to obtain a training sample set including a plurality of training samples, wherein the training samples include feature data of a mobile device for device user authentication and a label corresponding to the feature data;

[0088] The second module is used to train a preset LSTM model using the training sample set to obtain a certification model;

[0089] The characteristic data of the user of the mobile device to be authenticated is obtained as the input of the authentication model, and the output of the authentication model under the input is used as the authentication result of the device to be authenticated.

[0090] Preferably, the second module comprises:

[0091] The first submodule is used to learn the network parameters of the LSTM neural network using the enhanced SGD algorithm based on the training sample set.

[0092] The first component is used to randomly select samples from the training sample set to enter the group using a random sampling probability n / N, where n is the number of randomly selected training samples and N is the number of samples in the training sample set;

[0093] A second component is used to calculate the gradient value of the neural network using the sample entry group and a preset loss function;

[0094] A third component is used to clip the gradient value to obtain a clipped gradient value;

[0095] The fourth component is used to add Gaussian noise as a robust factor to the clipped gradient value to obtain a robust gradient value

[0096] The fifth component is used to perform gradient descent operation according to the following formula:

[0097]

[0098] Where t is the current iteration number, t = 0, 1, 2, ... T, T is the maximum iteration number, η t is the learning rate for the preset t-th iteration, t=0, 1, 2, ... T, θ t The network parameters of the tth iteration, θ0 is the preset initial network parameters, is the robust gradient value calculated for the tth iteration.

[0099] Preferably, the third subunit is used to use the sample to enter group G according to the following formula t And the preset loss function calculates the gradient value g t (x i ):

[0100]

[0101] in, is the preset loss function:

[0102]

[0103] Among them, x i is the i-th training sample in the sample entry group, θ represents the network parameters; y i Represents the training sample x i The corresponding labels are 0 for other people’s data and 1 for personal data; f(x i , θ) indicates that the input is the training sample x i The classification value calculated when the network parameter is θ has a value range of 0 to 1.

[0104] Preferably, the third component performs standard clipping on the gradient value g by L1 norm. t (x i ) is cut to obtain

[0105]

[0106] Use in formula Replace the original gradient vector g t (x i ), where C is the clipping threshold, which is used to control the gradient norm boundary. Represents the new gradient vector.

[0107] like Figure 1 As shown, the mobile device user authentication device constructed based on the optimized LSTM of the present invention includes a mobile device end and a server end. When performing mobile device authentication based on the mobile device user authentication device, the following steps are included:

[0108] S1. Data collection, obtaining training sample sets and transmitting them to the server: The mobile device collects a large amount of feature data for mobile device user authentication and the labels corresponding to each data.

[0109] In this embodiment, the characteristic data includes data such as an acceleration sensor, a gravity sensor, and a gyroscope of the mobile device that can identify user characteristics.

[0110] Usually, this special data is collected by the built-in sensor of the mobile device according to the preset sampling interval, and finally forms the corresponding time series data.

[0111] S2. The server performs denoising and normalization on each piece of collected feature data to obtain training data (ie, the training sample includes the processed feature data and the corresponding label). The processing in this embodiment mainly includes the processing of three major anomalies: equal value anomaly, jump point anomaly and zero value anomaly.

[0112] S3. Use the training data to learn the network parameters of the LSTM neural network to obtain a certification model (i.e., model).

[0113] This process is usually carried out on the server side, which can also be understood as being carried out online.

[0114] S4. Use the trained authentication model to perform device user authentication.

[0115] It should be noted that in step S4, the server can push the authentication model to the mobile device, and the device can use the model for offline authentication. It is also possible to upload the feature number used for user authentication to the server, and the server can authenticate the accuracy of the model.

[0116] In the step S3, the training data is sent to the LSTM neural network for learning to obtain the authentication model, which is mainly based on the stochastic gradient descent algorithm. Preferably, the enhanced SGD algorithm is used for gradient descent during training in this embodiment.

[0117] The structure of the LSTM neural network in this embodiment is as follows Figure 2 As shown in the figure, it mainly includes four layers, namely input layer, LSTM layer, classification layer and output layer. Among them, the input layer mainly inputs the preprocessed time series data. The input layer is a multi-dimensional vector (the length of the input layer is the dimension of the vector). The classification layer is softmax, and the output layer is 0 or 1. The LSTM layer consists of two layers of LSTM neurons, each layer has 32 neurons (LSTM cells), and the gradient descent algorithm is an enhanced SGD algorithm, which mainly includes pruning, grouping, robustness factor and tuning. The classification layer mainly uses Dense and Softmax for normalization and sorting. The final result is output by the output layer, and the output of the output layer is 0 or 1, indicating the support certification result.

[0118] Specifically, Figure 3 As shown, the model training using the enhanced SGD algorithm in this embodiment includes the following steps:

[0119] S301, determine input parameters and construct a loss function.

[0120] The input parameters include the determined training samples {x1, ..., x N}, learning rate η t The initial value of , the initial value of the noise ratio σ, which will be adjusted later, the fixed value of the group size G, the initial value of the gradient norm boundary C,

[0121] The loss function is as follows:

[0122]

[0123] Among them, x i is the i-th training sample, θ represents the network parameters; y i Represents the i-th training sample x i The corresponding labels are 0 for other people’s data and 1 for personal data; f(x i , θ) indicates that the input in the training process is the i-th training sample x i The classification value calculated when the network parameter is θ has a value range of 0 to 1.

[0124] The training samples in this implementation are composed of 9 dimensions of data in the x, y, and z axes of the acceleration, gyroscope, and gravity sensors. The acquisition time is 3 seconds, the acquisition frequency is 50 Hz, and each axis is 3 seconds for a total of 150 time series data. i is a vector (matrix) of 9*150. Correspondingly, the length of the input layer in the LSTM neural network of this embodiment is 150.

[0125] In this implementation, the range of the learning rate used is 0 to 1. Using 0.01 as the step size for optimization, it is found that the best range is 0 to 0.01, and further, the learning rate is preferably 0.0025.

[0126] The group size n can be 8, 16, 32, 64, 128, or 256, with 16 being preferred in this embodiment.

[0127] The initial value of the gradient norm boundary C can be 1 to 2. In this implementation, the gradient norm boundary C is 1.2.

[0128] S302: Initialize a random value θ0 as an initial network parameter, set a termination condition, perform an iterative operation to optimize the gradient value, and stop the iteration until the termination condition is reached, and use the gradient value obtained by the last iterative operation as the final gradient value.

[0129] In this embodiment, the termination condition is reaching a preset number of iterations T. Specifically: if t∈[T] (ie, t=0, 1, 2, . . . , T), an iteration operation is performed, and t represents the current number of iterations.

[0130] The following operations are performed at the tth iteration:

[0131] Step 1: Use random sampling probability n / N to select random samples from the training samples to enter group G t , n is the number of randomly selected training samples, and N represents the size of the number of samples in the training sample set.

[0132] Step 2: Calculate the gradient, for each i∈G t , calculate g t (x i ):

[0133]

[0134] Step 3: Standard pruning by L1 norm:

[0135]

[0136] Use in formula Replace the original gradient vector g t (x i ), where C is the clipping threshold, which controls the gradient norm boundary. Represents the new gradient vector.

[0137] Step 4: Add a robust factor (Gaussian noise) and add the time-dependent Gaussian noise to the gradient at the tth iteration using the following formula:

[0138]

[0139] in:

[0140]

[0141] The α in the formula is selected from {0.01, 0.3, 1.0}, preferably 0.01, and r is 0.55.

[0142] Step 4: Perform gradient descent operation:

[0143]

[0144] After the operation, return to step 3 and stop iterating until the condition t∈[T] is no longer satisfied.

[0145] Step 5: Output the final parameter θ T At this point, the enhanced SGD algorithm is complete.

[0146] Research has found that if Gaussian noise is added directly after the gradient is calculated, there will be a problem of the gradient changing too quickly. If Gaussian noise disturbance is added to a larger gradient, the final model will not be significantly affected, and the robustness of the model will be limited. In this application, the calculated gradient is first clipped. After clipping, the input gradient can be guaranteed to be within a certain stable range. The added noise can just offset the noise in the previous training data, thereby greatly increasing the robustness of the model and helping to improve the accuracy of recognition.

[0147] The present invention also provides a method for obtaining a training sample set, which is specifically based on an evolutionary game-based heuristic data collection method. The method overcomes the shortcomings of existing data collection methods, such as waste of resources and uncontrollable collection process during data collection. The present invention provides a data collection method based on an evolutionary game mechanism that continuously provides the best strategy. By generating revenue incentives, users are prompted to generate effective data for collection, and the collection strategy is continuously adjusted through user usage data to achieve Nash equilibrium.

[0148] A heuristic data collection method based on evolutionary game is implemented based on a heuristic data collection system. In the heuristic data collection system, the participating users and a server are connected via a network.

[0149] In one embodiment, a heuristic data collection method based on evolutionary game of the present invention performs the following operations on the server side:

[0150] Receiving an access request sent by a participating user to a server, wherein the access request includes a data collection sensing process selected from a preset data collection sensing process set;

[0151] Sending a response to the received access request to the participating user and allowing the participating user to access;

[0152] Receiving sensor data collected by the participating user according to the sensor process of data collection in the access request;

[0153] A strategy adjustment plan is formulated for the participating user based on the benefits calculated for the received sensor data and the status information of the participating user and the average benefits of all participating users of the heuristic data collection system, and sent to the participating user, wherein the strategy adjustment plan is used as a reference for the participating user to adjust the sensing process of data collection;

[0154] The status information of the participating users includes the participating users' own benefits and the average benefits of all participating users in the system.

[0155] The benefit calculated for the received sensor data is performed according to Φ:

[0156]

[0157] Among them, the set of sensing processes P, the profit calculation method, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed investment budget of sensing process p, N represents the number of users who choose sensing process p at time t, p∈P.

[0158] Furthermore, in this embodiment, the strategy adjustment scheme formulated for the participating users according to the status information of the participating users and the average income of all participating users of the heuristic data acquisition system includes:

[0159] For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval;

[0160] The difference δ is calculated according to the following formula * :

[0161]

[0162] Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the revenue of sensing process i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P];

[0163] According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

[0164] The success of a strategy is not determined solely by the quality of the strategy itself, but how good it is in a set of optional strategies and how often other strategies appear in a competitive group. In this embodiment, by setting the global strategy adaptation factor α, the strategy can be effectively adjusted, thereby improving the efficiency of data collection.

[0165] In reality, users are often rational and limited, and do not always maximize their interest tasks in actual use, which means that even if users find that their benefits are less than the average, they may stick to the current strategy and not change it. Based on the above assumptions, using the parameters in the random probability parameter (i.e., the random value ξ), participating users will change their strategies with this certain probability.

[0166] In another embodiment, a heuristic data collection method based on evolutionary game performs the following operations on the participating user end:

[0167] Sending an access request to a server, wherein the access request includes a data collection and sensing process selected from a preset data collection and sensing process set;

[0168] receiving a response from the server to the access request, and accessing the server;

[0169] Collect sensor data according to the data collection and sensing process in the access request and send it to the server;

[0170] The receiving server calculates the benefits of the received sensing data and formulates a strategy adjustment plan for the participating users according to the average benefits of all participating users of the heuristic data collection system, wherein the strategy adjustment plan is used as a reference for the participating users to adjust the data collection sensing process.

[0171] Another embodiment of the heuristic data collection method based on evolutionary game of the present invention is as follows: Figure 4 As shown, it is divided into the user side and the server side, including the following steps:

[0172] S401: A new user (ie, a participating user) joins the smartphone sensor collection process, selects a sensing process and confirms it, and informs the server what data will be collected.

[0173] S402: The effectiveness of user collection requires that the mobile phone screen is turned on and the mobile phone foreground application is switched. In addition, a sensing process includes multiple collection tasks, and each task can be completed by the user within a time slot. For example, if the ideal data collection duration is 3 seconds, if the user uses the mobile phone for less than 3 seconds (the mobile phone screen is turned off during the collection process), the collection is terminated. The user executes the application program in the root sensing process and obtains the sensor data and sends it to the server.

[0174] S403: The server receives the data and calculates the revenue, and sends the revenue to the participating users;

[0175] In step S403, the profit is calculated by method Φ.

[0176] Another embodiment of the heuristic data collection method based on evolutionary game of the present invention is as follows: Figure 5 As shown, the following steps are included:

[0177] S501: Data collection preparation, the number of smartphone users is U=1, 2, ..., U, and the size depends on the number of participating users; a limited set of data collection sensing processes P=1, 2, ..., P is preset in the server, and each sensing process contains sensor data collection tasks for different types of applications;

[0178] S502: A new user joins the sensor data collection process and can only join one sensor process, and submits the selection of what data to collect to the server.

[0179] S503: The user executes the application to obtain sensor data and sends it to the server. A sensing process will include multiple acquisition tasks, and each task can be completed by the user within a time slot (i.e., a fixed time period). The acquisition task requires the mobile phone screen to be turned on and the mobile phone foreground application to be switched.

[0180] S504: The server receives the sensor data, starts calculating the revenue, and returns the revenue to the user.

[0181] S505: The server will provide the user with the average revenue and participating user status information, and provide a notification of collection strategy adjustment based on the information, and the user can refer to it to change the data collection strategy. If the user confirms the selection, the new sensing process will be used for data collection and revenue calculation; if the user discards the selection, the original sensing process will still be used.

[0182] Optionally, the step of adjusting the collection strategy by the server through evolutionary game in S505 includes the following steps:

[0183] S5051: The benefit calculated for the received sensor data is performed according to Φ:

[0184]

[0185] Among them, the set of sensing processes P, the profit calculation method, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed investment budget of sensing process p, N represents the number of users who choose sensing process p at time t, p∈P.

[0186] S5052: Initialize the policy system, set the global policy adaptation factor α∈(0,1], and each user randomly selects a sensing process.

[0187] S5053: For each user u and each time slot t, perform the selected sensing process P u Required data collection.

[0188] S5054: Send the collected data to the server and calculate the benefits through the method Φ. And send the benefits, participating user status information θ(t) and average benefits Φ(θ(t)) to the participating users.

[0189] S5055: If the benefit received is less than the average benefit, or The system will reselect a random value ξ in the uniformly distributed (0, 1] interval.

[0190] In reality, users are often rational and limited, and do not always maximize their interest tasks in actual use, which means that even if users find that their benefits are less than the average, they may stick to the current strategy and not change the strategy. Based on the above assumptions, we use the parameters in the random probability parameters, and the participating users will change their strategies under this certain probability.

[0191] S5056: Calculate difference δ * :

[0192]

[0193] Where α is the global strategy adaptation factor, S i is the revenue of sensing process i, θ i (t) represents the proportion of users who choose the sensing process at time t. For all t, ∑ i∈P θ i (t)=1.

[0194] S5057: If δ * If δ is less than 0, the sensor process p′ is recommended to the user for selection. * If it is greater than 0, data collection is still performed according to the original sensing process, and the process returns to step S503.

[0195] In order to effectively collect data representing user behavior and improve data coverage during the training phase, the invention adopts an evolutionary stable participation game framework for effective data collection. It also generates revenue incentives to encourage users to generate effective data for collection, and continuously adjusts the collection strategy through user usage data to achieve a balance.

[0196] In summary, the present invention proposes a heuristic data collection method based on evolutionary game, which encourages users to participate in data collection through incentives and uses a participatory game mechanism to dynamically adjust the collection strategy to improve the efficiency of data collection and the effectiveness of data.

[0197] It should be clearly understood that the present invention describes how to form and use specific examples, but the principles of the present invention are not limited to any details of these examples. On the contrary, based on the teachings of the content disclosed in the present invention, these principles can be applied to many other embodiments.

[0198] The following is an embodiment of the device of the present invention, which can be used to implement the method of the present invention. In the following description of the system, the same parts as the above method will not be repeated.

[0199] The present invention also provides a heuristic data collection device based on evolutionary game, comprising:

[0200] The first unit is used to receive an access request sent by a participating user to a server, wherein the access request includes a data collection sensing process selected from a preset data collection sensing process set;

[0201] The second unit is used to send a response to the received access request to the participating user and allow the participating user to access;

[0202] A third unit is used to receive sensor data collected by the participating user according to the sensor process of data collection in the access request;

[0203] The fourth unit is used for calculating the benefits of the received sensor data, and formulating a strategy adjustment plan for the participating user according to the status information of the participating user and the average benefits of all participating users of the heuristic data collection system, and sending the strategy adjustment plan to the participating user, wherein the strategy adjustment plan is used as a reference for the participating user to adjust the sensing process of data collection;

[0204] The status information of the participating users includes the participating users' own benefits.

[0205] The fourth unit calculates the benefit of the received sensor data according to Φ:

[0206]

[0207] Among them, the set of sensing processes P, the profit calculation method, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed investment budget of sensing process p, N represents the number of users who choose sensing process p at time t, p∈P.

[0208] The fourth unit performs the following operations when formulating a strategy adjustment plan for the participating users according to the status information of the participating users and the average benefits of all participating users of the heuristic data acquisition system:

[0209] For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval;

[0210] The difference δ is calculated according to the following formula * :

[0211]

[0212] Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the revenue of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P];

[0213] According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

[0214] The present invention also provides a heuristic data collection device based on evolutionary game, comprising:

[0215] A fifth unit is used to send an access request to the server, wherein the access request includes a data collection and sensing process selected from a preset data collection and sensing process set;

[0216] A sixth unit is used to receive a response from the server to the access request and access the server;

[0217] A seventh unit is used to collect sensor data according to the data collection and sensing process in the access request and send it to the server;

[0218] The receiving server calculates the benefits of the received sensing data and formulates a strategy adjustment plan for the participating users according to the average benefits of all participating users of the heuristic data collection system, wherein the strategy adjustment plan is used as a reference for the participating users to adjust the data collection sensing process.

[0219] In another embodiment of the present invention, an electronic device includes:

[0220] one or more processors;

[0221] A storage device for storing one or more programs;

[0222] When the one or more programs are executed by the one or more processors, the one or more processors implement the above method.

[0223] In another embodiment of the present invention, a computer readable medium stores a computer program, which implements the above method when executed by a processor.

[0224] It should be understood that the computer device of the present embodiment includes a central processing unit (CPU), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage portion into a random access memory 1003. In the RAM, various programs and data required for the operation of the system 1000 are also stored. The CPU, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0225] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), the above-mentioned functions defined in the terminal of the present application are executed.

[0226] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0227] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0228] The units involved in the embodiments described in the present application may be implemented by software or hardware. The units or modules described may also be arranged in a processor, wherein the names of these modules do not constitute a limitation on the modules themselves in some cases.

[0229] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structure, configuration or implementation method described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. A heuristic data collection method based on evolutionary game, applied to a heuristic data collection system including participating users and a server, characterized in that: Perform the following operations on the server: Receiving an access request sent by a participating user to a server, wherein the access request includes a data collection sensing process selected from a preset data collection sensing process set; Sending a response to the received access request to the participating user and allowing the participating user to access; Receiving sensor data collected by the participating user according to the sensor process of data collection in the access request; A strategy adjustment plan is formulated for the participating user based on the benefits calculated for the received sensor data and the status information of the participating user and the average benefits of all participating users of the heuristic data collection system, and sent to the participating user, wherein the strategy adjustment plan is used as a reference for the participating user to adjust the sensing process of data collection; The status information of the participating users includes the participating users' own benefits; The benefit calculated for the received sensor data is performed based on Φ: Among them, the sensing process set P, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed input budget of the sensing process p, N represents the number of users who choose the sensing process p at time t, p∈P; The strategy adjustment schemes formulated for participating users according to the status information of participating users and the average benefits of all participating users of the heuristic data acquisition system include: For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval; The difference δ is calculated according to the following formula * : Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the benefit of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P]; According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

2. A heuristic data collection method based on evolutionary game, applied to a heuristic data collection system including participating users and a server, characterized in that: Perform the following operations on the participating user side: Sending an access request to a server, wherein the access request includes a data collection and sensing process selected from a preset data collection and sensing process set; receiving a response from the server to the access request, and accessing the server; Collect sensor data according to the data collection and sensing process in the access request and send it to the server; The server receives a policy adjustment plan formulated by the participating user, wherein the policy adjustment plan is used as a reference for the participating user to adjust the data collection and sensing process. The server formulates the policy adjustment plan using the following method: A benefit calculated for the received sensor data, and a strategy adjustment plan formulated for the participating user according to the state information of the participating user and the average benefit of all participating users according to the heuristic data acquisition system, wherein the state information of the participating user includes the participating user's own benefit; The benefit calculated for the received sensor data is performed based on Φ: Among them, the sensing process set P, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed input budget of the sensing process p, N represents the number of users who choose the sensing process p at time t, p∈P; The strategy adjustment schemes formulated for participating users according to the status information of participating users and the average benefits of all participating users of the heuristic data acquisition system include: For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval; The difference δ is calculated according to the following formula * : Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the benefit of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P]; According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

3. A heuristic data collection device based on evolutionary game, characterized in that: include: The first unit is used to receive an access request sent by a participating user to a server, wherein the access request includes a data collection sensing process selected from a preset data collection sensing process set; The second unit is used to send a response to the received access request to the participating user and allow the participating user to access; A third unit is used to receive sensor data collected by the participating user according to the sensor process of data collection in the access request; The fourth unit is used for calculating the benefits of the received sensor data, and formulating a strategy adjustment plan for the participating user according to the status information of the participating user and the average benefits of all participating users of the heuristic data collection system, and sending the strategy adjustment plan to the participating user, wherein the strategy adjustment plan is used as a reference for the participating user to adjust the sensing process of data collection; The status information of the participating users includes the participating users' own benefits; The fourth unit calculates the benefit of the received sensor data according to Φ: Among them, the sensing process set P, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed input budget of the sensing process p, N represents the number of users who choose the sensing process p at time t, p∈P; The fourth unit performs the following operations when formulating a strategy adjustment plan for the participating users according to the status information of the participating users and the average benefits of all participating users of the heuristic data acquisition system: For the sensor data received at time t, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval; The difference δ is calculated according to the following formula * : Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the benefit of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P]; According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

4. A heuristic data collection device based on evolutionary game, characterized in that: include: A fifth unit is used to send an access request to the server, wherein the access request includes a data collection and sensing process selected from a preset data collection and sensing process set; A sixth unit is used to receive a response from the server to the access request and access the server; A seventh unit is used to collect sensor data according to the data collection and sensing process in the access request and send it to the server; The server receives a policy adjustment plan formulated by the participating user, and the policy adjustment plan is used as a reference for the participating user to adjust the data collection and sensing process; the server formulates the policy adjustment plan by the following method: A benefit calculated for the received sensor data, and a strategy adjustment plan formulated for the participating user according to the state information of the participating user and the average benefit of all participating users according to the heuristic data acquisition system, wherein the state information of the participating user includes the participating user's own benefit; The benefit calculated for the received sensor data is performed based on Φ: Among them, the sensing process set P, θ p (t) represents the proportion of users who choose sensing process p at time t, B p represents the fixed input budget of the sensing process p, N represents the number of users who choose the sensing process p at time t, p∈P; The strategy adjustment schemes formulated for participating users according to the status information of participating users and the average benefits of all participating users of the heuristic data acquisition system include: For the sensor data received at time tz, when the benefit of the sensor data is less than the average benefit, a random value ξ is selected again in the uniformly distributed (0, 1] interval; The difference δ is calculated according to the following formula * : Among them, α is the preset global strategy adaptation factor, α∈(0,1], S i is the benefit of sensor type i, θ i (t) represents the proportion of users who choose sensing process i at time t, satisfying ∑ i∈P θ i (t) = 1, i∈[1, P]; According to the difference δ * :Generate a strategy adjustment plan: If δ * If δ is less than 0, new sensing process participants are recommended for selection; * If greater than 0, a new sensing process is not recommended.

5. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 2.

6. A mobile device user authentication method based on optimized LSTM, characterized in that: include: Acquire a training sample set including a plurality of training samples, wherein the training samples include feature data of a mobile device for device user authentication and a label corresponding to the feature data; Using the training sample set to train a pre-set LSTM model to obtain a certification model; Acquire feature data of the user of the mobile device to be authenticated as input of the authentication model, and use the output of the authentication model under the input as the authentication result of the device to be authenticated, The training sample set comprising a plurality of training samples is obtained by using any one of the methods in claims 1 to 2.