本說明書實施例的目的是提供一種風險預測和風險預測模型的訓練方法、裝置及電子設備,以提高風險預測模型的識別準確率。
為解決上述技術問題,本說明書實施例是這樣實現的:
第一態樣,提出了一種風險預測方法,包括:
從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
第二態樣,提出了一種風險預測模型的訓練方法,包括:
獲取賦能機構的使用者的私有資料和目標群體使用者的私有資料,其中,所述目標群體使用者為所述賦能機構和目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料,所述賦能機構的使用者包括所述目標群體使用者;
基於所述賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型;
基於所述目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
其中,所述第一風險預測模型和所述第二風險預測模型用於聯合識別使用者的風險等級。
第三態樣,提出了一種風險預測裝置,包括:
獲取單元,從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
第一預測單元,將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
第二預測單元,將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
第三預測單元,基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
第四態樣,提出了一種風險預測模型的訓練裝置,包括:
資料獲取單元,獲取賦能機構的使用者的私有資料和目標群體使用者的私有資料,其中,所述目標群體使用者為所述賦能機構和目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料,所述賦能機構的使用者包括所述目標群體使用者;
第一訓練單元,基於所述賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型;
第二訓練單元,基於所述目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
其中,所述第一風險預測模型和所述第二風險預測模型用於聯合識別使用者的風險等級。
第五態樣,提出了一種電子設備,該電子設備包括:
處理器;以及
被安排成儲存電腦可執行指令的記憶體,所述可執行指令在被執行時使所述處理器執行以下操作:
從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
第六態樣,提出了一種電腦可讀儲存媒體,所述電腦可讀儲存媒體儲存一個或多個程式,所述一個或多個程式當被包括多個應用程式的電子設備執行時,使得所述電子設備執行以下操作:
從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
第七態樣,提出了一種電子設備,包括:
處理器;以及
被安排成儲存電腦可執行指令的記憶體,所述可執行指令在被執行時使所述處理器執行以下操作:
獲取與目標使用者的使用者標識對應的私有資料;
將與所述目標使用者的使用者標識對應的私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將與所述目標使用者的使用者標識對應的私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
第八態樣,提出了一種電腦可讀儲存媒體,所述電腦可讀儲存媒體儲存一個或多個程式,所述一個或多個程式當被包括多個應用程式的電子設備執行時,使得所述電子設備執行以下操作:
獲取與目標使用者的使用者標識對應的私有資料;
將與所述目標使用者的使用者標識對應的私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將與所述目標使用者的使用者標識對應的私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
由以上本說明書實施例提供的技術方案可見,本說明書實施例方案至少具備如下一種技術效果:
本說明書提供的一種或多個實施例,能夠基於賦能機構的使用者的私有資料透過同構遷移訓練得到的第一風險預測模型,對目標使用者的風險等級進行預測;並能夠基於賦能機構和目標機構共同的目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,對目標使用者的風險等級進行二次預測,並結合這兩次預測結果來確定目標使用者的風險等級。由於充分利用了賦能機構的使用者的私有資料、以及賦能機構和目標機構共同的目標群體使用者的私有資料,訓練得到第一風險預測模型和第二風險預測模型聯合對目標使用者的風險等級進行預測,提高了風險等級的預測準確率。The purpose of the embodiments of this specification is to provide a risk prediction and risk prediction model training method, device and electronic device, so as to improve the recognition accuracy of the risk prediction model.
In order to solve the above-mentioned technical problems, the embodiments of this specification are implemented as follows:
In the first aspect, a risk prediction method is proposed, including:
Obtain the first private data corresponding to the user ID of the target user from the user database of the enabling organization, and obtain the first private data corresponding to the user ID of the target user from the user database of the target organization 2. Private information;
The first private data is input into the first risk level prediction model, and the first risk level of the target user is predicted to be obtained, and the first risk level prediction model is based on the private data of the user of the empowering institution through obtained by isomorphic transfer training;
Inputting the first private data and the second private data into a second risk level prediction model to predict the second risk level of the target user, and the second risk level prediction model is based on the use of target groups The private data and corresponding labels of the users are obtained through vertical federated learning training, and the labels corresponding to the private data of the target group users are the fitting errors corresponding to the target group users in the first risk prediction model;
predicting a risk level of the target user based on the first risk level and the second risk level;
Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution , and the private data of the target group users in the target institution.
In the second aspect, a training method for a risk prediction model is proposed, including:
Obtain private information of users of the enabling organization and private information of target group users, wherein the target group users are the common users of the enabling organization and the target organization, and the private information of the target group users Including the private data of the target group user in the enabling organization, and the private data of the target group user in the target organization, and the users of the enabling organization include the target group user;
Based on the private data of the user of the enabling organization, obtain the first risk prediction model through isomorphic transfer training;
Based on the private data of the target group users and the corresponding labels, a second risk prediction model is obtained through longitudinal federated learning training, and the labels corresponding to the private data of the target group users are the target group users in the first risk prediction model. The corresponding fitting error;
Wherein, the first risk prediction model and the second risk prediction model are used to jointly identify the user's risk level.
In a third aspect, a risk prediction device is proposed, including:
The acquiring unit acquires the first private data corresponding to the user identifier of the target user from the user database of the empowering institution, and acquires the user identifier corresponding to the target user from the user database of the target institution the corresponding second private information;
A first prediction unit, which inputs the first private data into a first risk level prediction model, and predicts the first risk level of the target user, and the first risk level prediction model is based on the use of the enabling agency The private data of the user is obtained through isomorphic transfer training;
The second prediction unit, which inputs the first private data and the second private data into a second risk level prediction model, and predicts the second risk level of the target user, and the second risk level prediction model It is obtained through vertical federated learning training based on the private data of the target group users and the corresponding labels. The label corresponding to the private data of the target group users is the fitting error corresponding to the first risk prediction model of the target group users. ;
a third predicting unit, predicting the risk level of the target user based on the first risk level and the second risk level;
Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution , and the private data of the target group users in the target institution.
In the fourth aspect, a training device for a risk prediction model is proposed, including:
A data acquisition unit for acquiring private data of users of the enabling organization and private data of users of the target group, wherein the users of the target group are the common users of the enabling organization and the target organization, and the target group uses The private data of the user includes the private data of the target group user in the enabling organization, and the private data of the target group user in the target organization, and the user of the enabling organization includes the target group user;
The first training unit, based on the private data of the user of the empowerment agency, obtains a first risk prediction model through isomorphic transfer training;
The second training unit obtains a second risk prediction model through vertical federated learning training based on the private data of the target group users and the corresponding labels, and the labels corresponding to the private data of the target group users are the target group users in the The fitting error corresponding to the first risk prediction model;
Wherein, the first risk prediction model and the second risk prediction model are used to jointly identify the user's risk level.
In a fifth aspect, an electronic device is proposed, and the electronic device includes:
processor; and
memory arranged to store computer-executable instructions which, when executed, cause the processor to:
Obtain the first private data corresponding to the user ID of the target user from the user database of the enabling organization, and obtain the first private data corresponding to the user ID of the target user from the user database of the target organization 2. Private information;
The first private data is input into the first risk level prediction model, and the first risk level of the target user is predicted to be obtained, and the first risk level prediction model is based on the private data of the user of the empowering institution through obtained by isomorphic transfer training;
Inputting the first private data and the second private data into a second risk level prediction model to predict the second risk level of the target user, and the second risk level prediction model is based on the use of target groups The private data and corresponding labels of the users are obtained through vertical federated learning training, and the labels corresponding to the private data of the target group users are the fitting errors corresponding to the target group users in the first risk prediction model;
predicting a risk level of the target user based on the first risk level and the second risk level;
Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution , and the private data of the target group users in the target institution.
In a sixth aspect, a computer-readable storage medium is proposed, the computer-readable storage medium stores one or more programs, the one or more programs, when executed by an electronic device including a plurality of application programs, cause all the The described electronic device performs the following actions:
Obtain the first private data corresponding to the user ID of the target user from the user database of the enabling organization, and obtain the first private data corresponding to the user ID of the target user from the user database of the target organization 2. Private information;
The first private data is input into the first risk level prediction model, and the first risk level of the target user is predicted to be obtained, and the first risk level prediction model is based on the private data of the user of the empowering institution through obtained by isomorphic transfer training;
Inputting the first private data and the second private data into a second risk level prediction model to predict the second risk level of the target user, and the second risk level prediction model is based on the use of target groups The private data and corresponding labels of the users are obtained through vertical federated learning training, and the labels corresponding to the private data of the target group users are the fitting errors corresponding to the target group users in the first risk prediction model;
predicting a risk level of the target user based on the first risk level and the second risk level;
Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution , and the private data of the target group users in the target institution.
In a seventh aspect, an electronic device is proposed, including:
processor; and
memory arranged to store computer-executable instructions which, when executed, cause the processor to:
Obtain private information corresponding to the user ID of the target user;
Input the private data corresponding to the user identification of the target user into the first risk level prediction model, and predict the first risk level of the target user, and the first risk level prediction model is based on empowerment The private data of the users of the institution are obtained through isomorphic transfer training;
Input the private data corresponding to the user identification of the target user into the second risk level prediction model, and predict the second risk level of the target user, and the second risk level prediction model is based on the target group. The user's private data and the corresponding label are obtained through vertical federated learning training, and the label corresponding to the target group user's private data is the fitting error corresponding to the target group user in the first risk prediction model;
predicting a risk level of the target user based on the first risk level and the second risk level;
Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution , and the private data of the target group users in the target institution.
In an eighth aspect, a computer-readable storage medium is proposed, the computer-readable storage medium stores one or more programs, the one or more programs, when executed by an electronic device including a plurality of application programs, cause all the The described electronic device performs the following actions:
Obtain private information corresponding to the user ID of the target user;
Input the private data corresponding to the user identification of the target user into the first risk level prediction model, and predict the first risk level of the target user, and the first risk level prediction model is based on empowerment The private data of the users of the institution are obtained through isomorphic transfer training;
Input the private data corresponding to the user identification of the target user into the second risk level prediction model, and predict the second risk level of the target user, and the second risk level prediction model is based on the target group. The user's private data and the corresponding label are obtained through vertical federated learning training, and the label corresponding to the target group user's private data is the fitting error corresponding to the target group user in the first risk prediction model;
predicting a risk level of the target user based on the first risk level and the second risk level;
Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution , and the private data of the target group users in the target institution.
It can be seen from the technical solutions provided by the above embodiments of this specification that the solutions of the embodiments of this specification have at least one of the following technical effects:
One or more embodiments provided in this specification can predict the risk level of a target user based on a first risk prediction model obtained through isomorphic transfer training based on the private data of the user of the enabling institution; The private data and corresponding labels of the target group users shared by the institution and the target institution are trained through longitudinal federated learning to obtain a second risk prediction model, which makes a secondary prediction of the risk level of the target user, and combines the two prediction results to get Determine the risk level of the target user. Because the private data of the users of the enabling institution and the private data of the target group users shared by the enabling institution and the target institution are fully utilized, the training results of the combination of the first risk prediction model and the second risk prediction model on the target users are obtained. The risk level is predicted, which improves the prediction accuracy of the risk level.
為使本說明書的目的、技術方案和優點更加清楚,下面將結合本說明書具體實施例及相應的圖式對本說明書中的技術方案進行清楚、完整地描述。顯然,所描述的實施例僅是本檔一部分實施例,而不是全部的實施例。基於本檔中的實施例,本領域普通技術人員在沒有做出創造性勞動前提下所獲得的所有其他實施例,都屬於本檔案保護的範圍。
以下結合圖式,詳細說明本說明書各實施例提供的技術方案。
為提高風險預測模型的識別準確率,本說明書一個或多個實施例提供一種風險預測方法,能夠基於賦能機構的使用者的私有資料透過同構遷移訓練得到的第一風險預測模型,對目標使用者的風險等級進行預測;並能夠基於賦能機構和目標機構共同的目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,對目標使用者的風險等級進行二次預測,並結合這兩次預測結果來確定目標使用者的風險等級。
由於充分利用了賦能機構的使用者的私有資料、以及賦能機構和目標機構共同的目標群體使用者的私有資料,訓練得到第一風險預測模型和第二風險預測模型聯合對目標使用者的風險等級進行預測,且第二風險預測模型在訓練時是以第一風險預測模型的擬合誤差為目標的,最後綜合第一風險預測模型和第二風險預測模型的預測結果,極大地提高了對使用者風險等級的預測準確率。
應理解,本說明書實施例提供的風險預測方法的執行主體,可以但不限於伺服器等能夠被配置為執行本說明書實施例提供的該方法裝置中的至少一種。
為便於描述,下文以該方法的執行主體為能夠執行該方法的伺服器為例,對該方法的實施方式進行介紹。可以理解,該方法的執行主體為伺服器只是一種示例性的說明,並不應理解為對該方法的限定。
圖1是本說明書的一個實施例提供的一種風險預測方法的實施流程示意圖。圖1的方法可包括:
S110,從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第二私有資料。
應理解,賦能機構與目標機構之間可以存在直接的合作關係,也可以存在間接的合作關係(即可以是透過中間機構建立的合作關係),且後文所述的第一風險預測模型和第二風險預測模型均是為目標機構服務的,而由於目標機構和賦能機構之間往往存在一些共同的使用者。在對這些使用者的風險等級進行預測時,則可以基於這些使用者的使用者標識,從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第二私有資料。
其中,第一私有資料和第二私有資料具體可包括目標使用者的交易資料資訊、目標使用者的身份資料資訊、目標使用者的帳號資料資訊、目標使用者的註冊資料資訊、目標使用者的職業、年齡、收入等等。
S120,將第一私有資料輸入到第一風險等級預測模型中,預測得到目標使用者的第一風險等級,第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的。
該第一風險等級具體可以是一個風險分值,取值範圍可以是[0, 1]。
S130,將第一私有資料和第二私有資料輸入到第二風險等級預測模型中,預測得到目標使用者的第二風險等級,第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差。
其中,目標群體使用者為賦能機構和目標機構的共同使用者,目標群體使用者的私有資料包括目標群體使用者在賦能機構中的私有資料、和目標群體使用者在目標機構中的私有資料。
需要說明的是,第二風險等級預測模型在訓練時具體可以是以第一風險等級預測模型的擬合誤差為預測目標訓練得到的。其中,第一風險等級預測模型的擬合誤差error=真實值Y-預測值Y1。
S140,基於第一風險等級和第二風險等級,預測目標使用者的風險等級。
可選地,為了更好地融合第一風險預測模型和第二風險預測模型的模型預測結果,本說明書一個或多個實施例可以透過加法模型來融合兩者的預測結果。具體地,基於第一風險等級和第二風險等級,確定目標使用者的風險等級,包括:
透過加法模型基於第一風險等級和第二風險等級,確定目標使用者的風險等級。
其中,加法模型由多個基模型相加而成,在本說明書實施例中,該加法模型由第一風險預測模型和第二風險預測模型相加而成,即預測值F(x)=f1(x)+f2(x),其中,f1(x)為第一風險預測模型的預測結果即第一風險等級,f2(x)為第二風險預測模型的預測結果即第二風險等級。具體來說,假設第一風險預測模型的預測目標是f1(x),那麼第二風險預測模型的預測目標是Y-f1(x),Y為真實值,那麼透過加法模型得到的預測值則是F(x)=f1(x)+f2(x)=Y,即預測值的範圍還是[0, 1]。
下面結合圖2所示的風險預測方法應用在一種實際場景中的示意圖,對風險預測方法的實施過程進行詳細描述,包括:
S21,基於賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型,其中,賦能機構的使用者包括一些目標機構的使用者。
應理解,為了充分利用賦能機構的使用者的私有資料,這裡所述的賦能機構的使用者的私有資料具體可以是該賦能機構的全量使用者的私有資料。
S22,從賦能機構的使用者資料庫中獲取與目標使用者的使用者ID相對應的第一私有資料,並將第一私有資料輸入到第一風險預測模型中,以透過第一風險預測模型對目標使用者的風險等級進行預測,輸出目標使用者的第一風險等級的預測值Y1。
S23,獲取第一風險預測模型的擬合誤差error,即目標使用者的真實風險等級值Y與Y1的差值error=Y-Y1。
S24,基於目標群體使用者的私有資料,並以第一風險預測模型的擬合誤差error為對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型。
其中,目標群體使用者為賦能機構和目標機構的共同使用者,目標群體使用者的私有資料包括目標群體使用者在賦能機構中的私有資料、和目標群體使用者在目標機構中的私有資料。且第二風險預測模型是以第一風險預測模型的擬合誤差error為預測目標訓練得到的。
S25,從目標機構的使用者資料庫中獲取與目標使用者的ID相對應的第二私有資料,並將S22獲取的第一私有資料和該第二私有資料輸入到第二風險預測模型中,以透過第二風險預測模型對目標使用者的風險等級進行預測,輸出目標使用者的第二風險等級的預測值Y2。
S26,基於加法模型得到目標使用者的風險等級,輸出得到目標使用者的風險等級的預測值Y1+Y2。
本說明書提供的一種或多個實施例,能夠基於賦能機構的使用者的私有資料透過同構遷移訓練得到的第一風險預測模型,對目標使用者的風險等級進行預測;並能夠基於賦能機構和目標機構共同的目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,對目標使用者的風險等級進行二次預測,並結合這兩次預測結果來確定目標使用者的風險等級。由於充分利用了賦能機構的使用者的私有資料、以及賦能機構和目標機構共同的目標群體使用者的私有資料,訓練得到第一風險預測模型和第二風險預測模型聯合對目標使用者的風險等級進行預測,提高了風險等級的預測準確率。
圖3是本說明書的一個實施例提供的一種風險預測模型的訓練方法的實施流程示意圖,包括:
S310,獲取賦能機構的使用者的私有資料和目標群體使用者的私有資料,其中,目標群體使用者為賦能機構和目標機構的共同使用者,目標群體使用者的私有資料包括目標群體使用者在賦能機構中的私有資料、和目標群體使用者在目標機構中的私有資料,賦能機構的使用者包括目標群體使用者。
其中,賦能機構期望透過在保護賦能機構本身和目標機構的私有資料的前提下,聯合使用賦能機構的使用者的私有資料和目標機構的私有資料,共同完成第一風險預測模型和第二風險預測模型的訓練。基於此,本說明書實施例採用同構遷移和縱向聯邦學習的模型訓練方式,分別訓練得到第一風險預測模型和第二風險預測模型,並將第一風險預測模型和第二風險預測模型聯合起來預測使用者的風險等級。
由於同構遷移只需使用賦能機構本身所有的使用者的私有資料,並結合賦能機構和目標機構共同的目標群體使用者在賦能機構中的私有資料,對目標機構進行了一次模型適配,得到第一風險預測模型;再透過縱向聯邦學習使用目標群體使用者在賦能機構中的私有資料和目標機構中的私有資料,訓練得到第二風險預測模型,充分利用了賦能機構所有的使用者的私有資料和目標機構可提供的目標群體使用者的私有資料,提高了風險預測的準確率。
如圖4所示,為本說明書實施例提供的透過同構遷移和縱向聯邦學習進行模型訓練的示意圖。在圖4(a)中,灰色區域為賦能機構擁有的所有使用者ID,以及對應的私有資料(即圖示的源域+目標域部分),目標域中包含的使用者的私有資料為賦能機構和目標機構的共同使用者在賦能機構中的私有資料,也就是說,目標域中包含的使用者的私有資料為賦能機構和目標機構重疊的那部分資料。
S320,基於賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型。
應理解,為了訓練得到適配於目標機構的第一風險預測模型,本說明書一個或多個實施例在透過同構遷移訓練得到第一風險預測模型時,應首先獲取賦能機構與目標機構的共同使用者即目標群體使用者在賦能機構中的私有資料。具體地,基於賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型,包括:
獲取賦能機構中目標群體使用者的私有資料;
基於賦能機構的使用者的私有資料和賦能機構中所述目標群體使用者的私有資料,透過同構遷移訓練得到第一風險預測模型。
如圖4(a)所示,為本說明書實施例提供的透過同構遷移訓練得到第一風險預測模型的示意圖,具體過程包括:
首先,用源域中包含的使用者的私有資料訓練得到神經網路類模型,具體訓練方式本說明書實施例對此不作具體限定;然後,對模型網路的每一層,計算源域中包含的使用者的私有資料在此層輸出的均值μ1和標準差σ1,以及目標域中包含的使用者的私有資料在此層輸出的均值μ2和標準差σ2;再透過訓練得到的模型對目標域中包含的使用者的私有資料進行預測,得到預測值U,再對該預測值進行統一資料分佈,得到預測值[(U-μ2)/σ2]*σ1+μ1,從而統一訓練得到的第一風險預測模型對源域和目標域中的使用者的私有資料的預測結果的範圍。
S330,基於目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差。
其中,第一風險預測模型和第二風險預測模型用於聯合識別使用者的風險等級。
應理解,為了提高風險預測模型的預測準確率,從而更好地為目標機構服務,本說明書一個或多個實施例還可透過縱向聯邦學習訓練得到第二風險預測模型。具體地,基於目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,包括:
基於第一風險預測模型對測試資料的預測值和測試資料對應的真實值,獲取第一風險預測模型的擬合誤差;
基於目標群體使用者的私有資料,透過縱向聯邦學習訓練得到第二風險預測模型,直到第二風險預測模型的預測值逼近第一風險預測模型的擬合誤差。
如圖4(b)所示,為本說明書實施例提供的透過縱向聯邦學習訓練得到第二風險預測模型的示意圖。該第二風險預測模型具體是基於賦能機構和目標機構的共同使用者,即目標群體使用者在賦能機構中的私有資料和目標群體使用者在目標機構中的私有資料,以賦能機構無法獲知或反推目標機構的資料,且目標機構無法獲知或反推賦能機構的資料的前提下,透過縱向聯邦學習訓練得到的。
本說明書提供的一種或多個實施例,能夠基於賦能機構的使用者的私有資料透過同構遷移訓練得到的第一風險預測模型,對使用者的風險等級進行預測;並能夠基於賦能機構和目標機構共同的目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,對使用者的風險等級進行二次預測,並結合這兩次預測結果來確定使用者的風險等級。由於充分利用了賦能機構的使用者的私有資料、以及賦能機構和目標機構共同的目標群體使用者的私有資料,訓練得到第一風險預測模型和第二風險預測模型聯合對使用者的風險等級進行預測,提高了風險等級的預測準確率。
圖5是本說明書的一個實施例提供的一種風險預測裝置500的結構示意圖。請參考圖5,在一種軟體實施方式中,風險預測裝置500可包括:
獲取單元501,從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
第一預測單元502,將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
第二預測單元503,將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
第三預測單元504,基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
可選地,在一種實施方式中,所述第三預測單元504,用於:
透過加法模型基於所述第一風險等級和所述第二風險等級,確定所述目標使用者的風險等級。
風險預測裝置500能夠實現圖1~圖2的方法實施例的方法,具體可參考圖1~圖2所示實施例的風險預測方法,不再贅述。
圖6是本說明書的一個實施例提供的一種風險預測模型的訓練裝置600的結構示意圖。請參考圖6,在一種軟體實施方式中,風險預測模型的訓練裝置600可包括:
資料獲取單元601,獲取賦能機構的使用者的私有資料和目標群體使用者的私有資料,其中,所述目標群體使用者為所述賦能機構和目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料,所述賦能機構的使用者包括所述目標群體使用者;
第一訓練單元602,基於所述賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型;
第二訓練單元603,基於所述目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
其中,所述第一風險預測模型和所述第二風險預測模型用於聯合識別使用者的風險等級。
可選地,在一種實施方式中,所述第一訓練單元602,用於:
獲取所述賦能機構中所述目標群體使用者的私有資料;
基於所述賦能機構的使用者的私有資料和所述賦能機構中所述目標群體使用者的私有資料,透過同構遷移訓練得到所述第一風險預測模型。
可選地,在一種實施方式中,所述第二訓練單元603,用於:
基於所述第一風險預測模型對測試資料的預測值和所述測試資料對應的真實值,獲取所述第一風險預測模型的擬合誤差;
基於所述目標群體使用者的私有資料,透過縱向聯邦學習訓練得到第二風險預測模型,直到所述第二風險預測模型的預測值逼近所述第一風險預測模型的擬合誤差。
風險預測模型的訓練裝置600能夠實現圖3~圖4的方法實施例的方法,具體可參考圖3~圖4所示實施例的風險預測模型的訓練方法,不再贅述。
圖7是本說明書的一個實施例電子設備的結構示意圖。請參考圖7,在硬體層面,該電子設備包括處理器,可選地還包括內部匯流排、網路介面、記憶體。其中,記憶體可能包含內部記憶體,例如高速隨機存取記憶體(Random-Access Memory,RAM),也可能還包括非揮發性記憶體(non-volatile memory),例如至少1個磁碟記憶體等。當然,該電子設備還可能包括其他業務所需要的硬體。
處理器、網路介面和記憶體可以透過內部匯流排相互連接,該內部匯流排可以是ISA(Industry Standard Architecture,工業標準架構)匯流排、PCI(Peripheral Component Interconnect,週邊組件互連標準)匯流排或EISA(Extended Industry Standard Architecture,延伸工業標準架構)匯流排等。所述匯流排可以分為位址匯流排、資料匯流排、控制匯流排等。為便於表示,圖7中僅用一個雙向箭頭表示,但並不表示僅有一根匯流排或一種類型的匯流排。
記憶體,用於存放程式。具體地,程式可以包括程式碼,所述程式碼包括電腦操作指令。記憶體可以包括內部記憶體和非揮發性記憶體,並向處理器提供指令和資料。
處理器從非揮發性記憶體中讀取對應的電腦程式到內部記憶體中然後運行,在邏輯層面上形成風險預測裝置。處理器,執行記憶體所存放的程式,並具體用於執行以下操作:
從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
上述如本說明書圖1~圖2所示實施例揭示的風險預測裝置執行的方法可以應用於處理器中,或者由處理器實現。處理器可能是一種積體電路晶片,具有訊號的處理能力。在實現過程中,上述方法的各步驟可以透過處理器中的硬體的整合邏輯電路或者軟體形式的指令完成。上述的處理器可以是通用處理器,包括中央處理器(Central Processing Unit,CPU)、網路處理器(Network Processor,NP)等;還可以是數位訊號處理器(Digital Signal Processor,DSP)、專用積體電路(Application Specific Integrated Circuit,ASIC)、現場可程式設計閘陣列(Field-Programmable Gate Array,FPGA)或者其他可程式設計邏輯裝置、分立閘或者電晶體邏輯裝置、分立硬體元件。可以實現或者執行本說明書實施例中的揭露的各方法、步驟及邏輯方塊圖。通用處理器可以是微處理器或者該處理器也可以是任何常規的處理器等。結合本說明書實施例所揭露的方法的步驟可以直接體現為硬體解碼處理器執行完成,或者用解碼處理器中的硬體及軟體模組組合執行完成。軟體模組可以位於隨機記憶體,快閃記憶體、唯讀記憶體,可程式設計唯讀記憶體或者電可讀寫可程式設計記憶體、暫存器等本領域成熟的儲存媒體中。該儲存媒體位於記憶體,處理器讀取記憶體中的資訊,結合其硬體完成上述方法的步驟。
該電子設備還可執行圖1~圖2的方法,並實現風險預測裝置在圖1~圖2所示實施例的功能,本說明書實施例在此不再贅述。
本說明書實施例還提出了一種電腦可讀儲存媒體,該電腦可讀儲存媒體儲存一個或多個程式,該一個或多個程式包括指令,該指令當被包括多個應用程式的可攜式電子設備執行時,能夠使該可攜式電子設備執行圖1~圖2所示實施例的方法,並具體用於執行以下操作:
從賦能機構的使用者資料庫中獲取與目標使用者的使用者標識對應的第一私有資料、以及從目標機構的使用者資料庫中獲取與所述目標使用者的使用者標識對應的第二私有資料;
將所述第一私有資料輸入到第一風險等級預測模型中,預測得到所述目標使用者的第一風險等級,所述第一風險等級預測模型為基於賦能機構的使用者的私有資料透過同構遷移訓練得到的;
將所述第一私有資料和所述第二私有資料輸入到第二風險等級預測模型中,預測得到所述目標使用者的第二風險等級,所述第二風險等級預測模型為基於目標群體使用者的私有資料及對應的標籤、透過縱向聯邦學習訓練得到的,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
基於所述第一風險等級和所述第二風險等級,預測所述目標使用者的風險等級;
其中,所述目標群體使用者為所述賦能機構和所述目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料。
當然,除了軟體實現方式之外,本說明書的電子設備並不排除其他實現方式,比如邏輯裝置抑或軟硬體結合的方式等等,也就是說以下處理流程的執行主體並不限定於各個邏輯單元,也可以是硬體或邏輯裝置。
圖8是本說明書的一個實施例電子設備的結構示意圖。請參考圖8,在硬體層面,該電子設備包括處理器,可選地還包括內部匯流排、網路介面、記憶體。其中,記憶體可能包含內部記憶體,例如高速隨機存取記憶體(Random-Access Memory,RAM),也可能還包括非揮發性記憶體(non-volatile memory),例如至少1個磁碟記憶體等。當然,該電子設備還可能包括其他業務所需要的硬體。
處理器、網路介面和記憶體可以透過內部匯流排相互連接,該內部匯流排可以是ISA(Industry Standard Architecture,工業標準架構)匯流排、PCI(Peripheral Component Interconnect,週邊組件互連標準)匯流排或EISA(Extended Industry Standard Architecture,延伸工業標準架構)匯流排等。所述匯流排可以分為位址匯流排、資料匯流排、控制匯流排等。為便於表示,圖8中僅用一個雙向箭頭表示,但並不表示僅有一根匯流排或一種類型的匯流排。
記憶體,用於存放程式。具體地,程式可以包括程式碼,所述程式碼包括電腦操作指令。記憶體可以包括內部記憶體和非揮發性記憶體,並向處理器提供指令和資料。
處理器從非揮發性記憶體中讀取對應的電腦程式到內部記憶體中然後運行,在邏輯層面上形成風險預測模型的訓練裝置。處理器,執行記憶體所存放的程式,並具體用於執行以下操作:
獲取賦能機構的使用者的私有資料和目標群體使用者的私有資料,其中,所述目標群體使用者為所述賦能機構和目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料,所述賦能機構的使用者包括所述目標群體使用者;
基於所述賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型;
基於所述目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
其中,所述第一風險預測模型和所述第二風險預測模型用於聯合識別使用者的風險等級。
上述如本說明書圖3~圖4所示實施例揭示的風險預測模型的訓練裝置執行的方法可以應用於處理器中,或者由處理器實現。處理器可能是一種積體電路晶片,具有訊號的處理能力。在實現過程中,上述方法的各步驟可以透過處理器中的硬體的整合邏輯電路或者軟體形式的指令完成。上述的處理器可以是通用處理器,包括中央處理器(Central Processing Unit,CPU)、網路處理器(Network Processor,NP)等;還可以是數位訊號處理器(Digital Signal Processor,DSP)、專用積體電路(Application Specific Integrated Circuit,ASIC)、現場可程式設計閘陣列(Field-Programmable Gate Array,FPGA)或者其他可程式設計邏輯裝置、分立閘或者電晶體邏輯裝置、分立硬體元件。可以實現或者執行本說明書實施例中的揭露的各方法、步驟及邏輯方塊圖。通用處理器可以是微處理器或者該處理器也可以是任何常規的處理器等。結合本說明書實施例所揭露的方法的步驟可以直接體現為硬體解碼處理器執行完成,或者用解碼處理器中的硬體及軟體模組組合執行完成。軟體模組可以位於隨機記憶體,快閃記憶體、唯讀記憶體,可程式設計唯讀記憶體或者電可讀寫可程式設計記憶體、暫存器等本領域成熟的儲存媒體中。該儲存媒體位於記憶體,處理器讀取記憶體中的資訊,結合其硬體完成上述方法的步驟。
該電子設備還可執行圖3~圖4的方法,並實現風險預測模型的訓練裝置在圖3~圖4所示實施例的功能,本說明書實施例在此不再贅述。
本說明書實施例還提出了一種電腦可讀儲存媒體,該電腦可讀儲存媒體儲存一個或多個程式,該一個或多個程式包括指令,該指令當被包括多個應用程式的可攜式電子設備執行時,能夠使該可攜式電子設備執行圖3~圖4所示實施例的方法,並具體用於執行以下操作:
獲取賦能機構的使用者的私有資料和目標群體使用者的私有資料,其中,所述目標群體使用者為所述賦能機構和目標機構的共同使用者,所述目標群體使用者的私有資料包括所述目標群體使用者在所述賦能機構中的私有資料、和所述目標群體使用者在目標機構中的私有資料,所述賦能機構的使用者包括所述目標群體使用者;
基於所述賦能機構的使用者的私有資料,透過同構遷移訓練得到第一風險預測模型;
基於所述目標群體使用者的私有資料及對應的標籤,透過縱向聯邦學習訓練得到第二風險預測模型,所述目標群體使用者的私有資料對應的標籤為目標群體使用者在第一風險預測模型對應的擬合誤差;
其中,所述第一風險預測模型和所述第二風險預測模型用於聯合識別使用者的風險等級。
上述對本說明書特定實施例進行了描述。其它實施例在所附申請專利範圍的範圍內。在一些情況下,在申請專利範圍中記載的動作或步驟可以按照不同於實施例中的順序來執行並且仍然可以實現期望的結果。另外,在圖式中描繪的過程不一定要求示出的特定順序或者連續順序才能實現期望的結果。在某些實施方式中,多工處理和並行處理也是可以的或者可能是有利的。
總之,以上所述僅為本說明書的較佳實施例而已,並非用於限定本說明書的保護範圍。凡在本說明書的精神和原則之內,所作的任何修改、等同替換、改進等,均應包含在本說明書的保護範圍之內。
上述實施例闡明的系統、裝置、模組或單元,具體可以由電腦晶片或實體實現,或者由具有某種功能的產品來實現。一種典型的實現設備為電腦。具體的,電腦例如可以為個人電腦、膝上型電腦、蜂巢式電話、相機電話、智慧型電話、個人數位助理、媒體播放機、導航設備、電子郵件設備、遊戲主機、平板電腦、可穿戴設備或者這些設備中的任何設備的組合。
電腦可讀媒體包括永久性和非永久性、可移動和非可移動媒體可以由任何方法或技術來實現資訊儲存。資訊可以是電腦可讀指令、資料結構、程式的模組或其他資料。電腦的儲存媒體的例子包括,但不限於相變內部記憶體(PRAM)、靜態隨機存取記憶體(SRAM)、動態隨機存取記憶體(DRAM)、其他類型的隨機存取記憶體(RAM)、唯讀記憶體(ROM)、電可抹除可程式設計唯讀記憶體(EEPROM)、快閃記憶體或其他內部記憶體技術、唯讀光碟唯讀記憶體(CD-ROM)、數位多功能光碟(DVD)或其他光學儲存器、磁盒式磁帶,磁帶式磁碟儲存器或其他磁性儲存裝置或任何其他非傳輸媒體,可用於儲存可以被計算設備存取的資訊。按照本文中的界定,電腦可讀媒體不包括暫態媒體(transitory media),如調變的資料訊號和載波。
還需要說明的是,術語“包括”、“包含”或者其任何其他變體意在涵蓋非排他性的包含,從而使得包括一系列要素的過程、方法、商品或者設備不僅包括那些要素,而且還包括沒有明確列出的其他要素,或者是還包括為這種過程、方法、商品或者設備所固有的要素。在沒有更多限制的情況下,由語句“包括一個……”限定的要素,並不排除在包括所述要素的過程、方法、商品或者設備中還存在另外的相同要素。
本說明書中的各個實施例均採用遞進的方式描述,各個實施例之間相同相似的部分互相參見即可,每個實施例重點說明的都是與其他實施例的不同之處。尤其,對於系統實施例而言,由於其基本相似於方法實施例,所以描述的比較簡單,相關之處參見方法實施例的部分說明即可。 In order to make the purpose, technical solutions and advantages of this specification clearer, the technical solutions in this specification will be clearly and completely described below with reference to specific embodiments of this specification and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this document, but not all of the embodiments. Based on the embodiments in this file, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of this file. The technical solutions provided by the embodiments of the present specification will be described in detail below with reference to the drawings. In order to improve the identification accuracy of the risk prediction model, one or more embodiments of the present specification provide a risk prediction method, which can analyze the target risk based on the first risk prediction model obtained by isomorphic transfer training based on the private data of the user of the enabling institution. The risk level of the user can be predicted; and based on the private data and corresponding labels of the target group users shared by the enabling agency and the target agency, a second risk prediction model can be obtained through vertical federated learning training, and the risk level of the target user can be calculated. Carry out a second prediction, and combine the results of the two predictions to determine the risk level of the target user. Because the private data of the users of the enabling institution and the private data of the target group users shared by the enabling institution and the target institution are fully utilized, the training results of the combination of the first risk prediction model and the second risk prediction model on the target users are obtained. The risk level is predicted, and the second risk prediction model is trained with the fitting error of the first risk prediction model as the goal. Finally, the prediction results of the first risk prediction model and the second risk prediction model are integrated, which greatly improves the performance. The prediction accuracy of the user's risk level. It should be understood that the execution body of the risk prediction method provided by the embodiments of the present specification may be, but not limited to, a server and the like that can be configured to execute at least one of the method and apparatus provided by the embodiments of the present specification. For the convenience of description, the implementation of the method is described below by taking an example that the execution body of the method is a server capable of executing the method. It can be understood that the execution body of the method is a server, which is only an exemplary description, and should not be construed as a limitation of the method. FIG. 1 is a schematic diagram of an implementation flow of a risk prediction method provided by an embodiment of the present specification. The method of FIG. 1 may include: S110: Acquire the first private data corresponding to the user ID of the target user from the user database of the enabling organization, and obtain the first private data corresponding to the target user from the user database of the target organization The second private data corresponding to the user ID of . It should be understood that there may be a direct cooperative relationship or an indirect cooperative relationship (that is, a cooperative relationship established through an intermediary agency) between the enabling agency and the target agency, and the first risk prediction model described later and The second risk prediction models are all for the target institution, and because there are often some common users between the target institution and the enabling institution. When predicting the risk levels of these users, based on the user identifiers of these users, the first private data corresponding to the user identifiers of the target users can be obtained from the user database of the enabling institution, and Obtain the second private data corresponding to the user ID of the target user from the user database of the target institution. The first private data and the second private data may specifically include transaction data information of the target user, identity data information of the target user, account data information of the target user, registration data information of the target user, Occupation, age, income, etc. S120: Input the first private data into the first risk level prediction model to predict the first risk level of the target user, and the first risk level prediction model is trained by isomorphic migration based on the private data of the user of the enabling organization owned. The first risk level may specifically be a risk score, and the value range may be [0, 1]. S130: Input the first private data and the second private data into the second risk level prediction model, and predict the second risk level of the target user, and the second risk level prediction model is based on the private data of the target group users and the corresponding The label is obtained through longitudinal federated learning training, and the label corresponding to the private data of the target group user is the fitting error corresponding to the target group user in the first risk prediction model. Among them, the target group users are the common users of the empowerment organization and the target organization, and the private data of the target group users include the private data of the target group users in the empowerment organization, and the private data of the target group users in the target organization. material. It should be noted that, during training, the second risk level prediction model may be specifically obtained by training the fitting error of the first risk level prediction model as the prediction target. Wherein, the fitting error error of the first risk level prediction model=true value Y-predicted value Y1. S140, predict the risk level of the target user based on the first risk level and the second risk level. Optionally, in order to better fuse the model prediction results of the first risk prediction model and the second risk prediction model, one or more embodiments of the present specification may use an additive model to fuse the prediction results of the two. Specifically, determining the risk level of the target user based on the first risk level and the second risk level includes: determining the risk level of the target user based on the first risk level and the second risk level through an additive model. The additive model is formed by adding a plurality of base models. In the embodiment of this specification, the additive model is formed by adding the first risk prediction model and the second risk prediction model, that is, the predicted value F(x)=f1 (x)+f2(x), where f1(x) is the prediction result of the first risk prediction model, that is, the first risk level, and f2(x) is the prediction result of the second risk prediction model, that is, the second risk level. Specifically, assuming that the prediction target of the first risk prediction model is f1(x), then the prediction target of the second risk prediction model is Y-f1(x), and Y is the true value, then the predicted value obtained through the additive model is Is F(x)=f1(x)+f2(x)=Y, that is, the range of predicted values is still [0, 1]. The following describes the implementation process of the risk prediction method in detail with reference to the schematic diagram of the application of the risk prediction method shown in Figure 2 in an actual scenario, including: S21, based on the private data of the user of the enabling organization, through the isomorphic transfer training A first risk prediction model is obtained, wherein the users of the enabling institution include some users of the target institution. It should be understood that, in order to make full use of the private information of the users of the enabling institution, the private information of the users of the enabling institution described here may specifically be the private information of all users of the enabling institution. S22: Obtain the first private data corresponding to the user ID of the target user from the user database of the enabling organization, and input the first private data into the first risk prediction model, so as to predict the risks through the first risk prediction model. The model predicts the risk level of the target user, and outputs the predicted value Y1 of the first risk level of the target user. S23: Obtain the fitting error error of the first risk prediction model, that is, the difference error=Y-Y1 between the real risk level value Y of the target user and Y1. S24, based on the private data of the target group users, and using the fitting error error of the first risk prediction model as a corresponding label, obtain a second risk prediction model through longitudinal federated learning training. Among them, the target group users are the common users of the empowerment organization and the target organization, and the private data of the target group users include the private data of the target group users in the empowerment organization, and the private data of the target group users in the target organization. material. And the second risk prediction model is obtained by training the fitting error error of the first risk prediction model as a prediction target. S25, obtain the second private data corresponding to the ID of the target user from the user database of the target institution, and input the first private data and the second private data obtained in S22 into the second risk prediction model, The target user's risk level is predicted through the second risk prediction model, and the predicted value Y2 of the target user's second risk level is output. S26, obtain the risk level of the target user based on the addition model, and output the predicted value Y1+Y2 of the risk level of the target user. One or more embodiments provided in this specification can predict the risk level of a target user based on a first risk prediction model obtained through isomorphic transfer training based on the private data of the user of the enabling institution; The private data and corresponding labels of the target group users shared by the institution and the target institution are trained through longitudinal federated learning to obtain a second risk prediction model, which makes a secondary prediction of the risk level of the target user, and combines the two prediction results to get Determine the risk level of the target user. Because the private data of the users of the enabling institution and the private data of the target group users shared by the enabling institution and the target institution are fully utilized, the training results of the combination of the first risk prediction model and the second risk prediction model on the target users are obtained. The risk level is predicted, which improves the prediction accuracy of the risk level. Fig. 3 is a schematic diagram of an implementation flow of a training method for a risk prediction model provided by an embodiment of the present specification, including: S310: Acquire private information of users of the enabling institution and private information of users of target groups, wherein the target group The user is the common user of the enabling institution and the target institution. The private information of the target group user includes the private information of the target group user in the enabling institution and the private information of the target group user in the target institution. Institutional users include target group users. Among them, the enabling agency expects to jointly complete the first risk prediction model and the third 2. Training of risk prediction models. Based on this, the embodiment of this specification adopts the model training methods of isomorphic transfer and vertical federated learning, respectively trains the first risk prediction model and the second risk prediction model, and combines the first risk prediction model and the second risk prediction model Predict the user's risk level. Since the isomorphic migration only needs to use the private data of all users of the enabling institution itself, and combine the private data of the target group users in the enabling institution shared by the enabling institution and the target institution, a model adaptation is carried out on the target institution. Then, through vertical federated learning, the private data of the target group users in the enabling institution and the private data of the target institution are used to train the second risk prediction model, which makes full use of all the enabling institutions. The private data of the users of the target group and the private data of the target group users provided by the target organization improve the accuracy of risk prediction. As shown in FIG. 4 , it is a schematic diagram of model training through isomorphic transfer and vertical federated learning provided by an embodiment of the present specification. In Figure 4(a), the gray area is all user IDs owned by the enabling organization and the corresponding private data (ie, the source domain + target domain part shown in the figure). The private data of the user contained in the target domain is The private data of the common users of the enabling institution and the target institution in the enabling institution, that is to say, the private data of the user contained in the target domain is the part of the overlapping data of the enabling institution and the target institution. S320 , obtaining a first risk prediction model through isomorphic transfer training based on the private data of the user of the enabling organization. It should be understood that, in order to obtain the first risk prediction model adapted to the target institution through training, when obtaining the first risk prediction model through isomorphic transfer training in one or more embodiments of this specification, the first risk prediction model of the enabling institution and the target institution should be obtained first. Co-users are private data of target group users in the enabling organization. Specifically, based on the private data of the users of the enabling institution, the first risk prediction model is obtained through isomorphic transfer training, including: obtaining the private data of the users of the target group in the enabling institution; based on the private data of the users of the enabling institution The data and the private data of the target group users in the enabling agency are trained through isomorphic transfer to obtain the first risk prediction model. As shown in FIG. 4( a ), a schematic diagram of obtaining a first risk prediction model through isomorphic transfer training provided in the embodiment of this specification, the specific process includes: First, using the private data of the user included in the source domain to train the neural network Network model, the specific training method is not specifically limited in the embodiment of this specification; then, for each layer of the model network, calculate the mean value μ1 and standard deviation σ1 output by the private data of the user contained in the source domain at this layer , and the output mean μ2 and standard deviation σ2 of the user’s private data contained in the target domain in this layer; then predict the private data of the user contained in the target domain through the model obtained by training to obtain the predicted value U, and then Perform a unified data distribution on the predicted value to obtain the predicted value [(U-μ2)/σ2]*σ1+μ1, so that the first risk prediction model obtained by unified training has the effect on the private data of users in the source domain and the target domain. The range of forecast results. S330 , a second risk prediction model is obtained through vertical federated learning training based on the private data of the target group users and the corresponding labels, and the labels corresponding to the private data of the target group users are the corresponding labels of the target group users in the first risk prediction model Fitting error. Wherein, the first risk prediction model and the second risk prediction model are used to jointly identify the user's risk level. It should be understood that, in order to improve the prediction accuracy of the risk prediction model, so as to better serve the target institution, one or more embodiments of the present specification may also obtain a second risk prediction model through longitudinal federated learning training. Specifically, based on the private data and corresponding labels of the target group users, a second risk prediction model is obtained through vertical federated learning training, including: the predicted value of the test data and the actual value corresponding to the test data based on the first risk prediction model, Obtain the fitting error of the first risk prediction model; Based on the private data of the target group users, obtain the second risk prediction model through longitudinal federated learning training, until the predicted value of the second risk prediction model is close to the fitting of the first risk prediction model error. As shown in FIG. 4( b ), a schematic diagram of obtaining a second risk prediction model through longitudinal federated learning training provided by the embodiment of the present specification. The second risk prediction model is specifically based on the common users of the enabling institution and the target institution, that is, the private data of the target group users in the enabling institution and the private information of the target group users in the target institution, so as to enable the institution to Obtained through vertical federated learning training on the premise that the information of the target institution cannot be obtained or reversed, and the target institution cannot know or reverse the information of the enabling institution. One or more embodiments provided in this specification can predict the user's risk level based on the first risk prediction model obtained through isomorphic transfer training based on the private data of the user of the enabling institution; and can predict the risk level of the user based on the enabling institution The private information and corresponding labels of the target group users shared with the target institution are trained through longitudinal federated learning to obtain a second risk prediction model, and the user's risk level is predicted twice, and the two prediction results are combined to determine the use of the risk level of the person. Because the private data of the users of the enabling institution and the private data of the target group users shared by the enabling institution and the target institution are fully utilized, the risk to the users of the combination of the first risk prediction model and the second risk prediction model is obtained by training. The prediction of the risk level improves the prediction accuracy of the risk level. FIG. 5 is a schematic structural diagram of a risk prediction apparatus 500 provided by an embodiment of the present specification. Referring to FIG. 5 , in a software implementation, the risk prediction apparatus 500 may include: an acquiring unit 501 for acquiring first private data corresponding to the user ID of the target user from the user database of the enabling institution, and Obtain the second private data corresponding to the user ID of the target user from the user database of the target institution; the first prediction unit 502 inputs the first private data into the first risk level prediction model, Predicting and obtaining the first risk level of the target user, and the first risk level prediction model is obtained through isomorphic transfer training based on the private data of the user of the enabling organization; the second predicting unit 503 is to convert the first risk level A private data and the second private data are input into the second risk level prediction model, and the second risk level of the target user is predicted to be obtained, and the second risk level prediction model is based on the private data of the target group users and the corresponding label, obtained through vertical federated learning training, the label corresponding to the private data of the target group user is the fitting error corresponding to the target group user in the first risk prediction model; the third prediction unit 504, based on the The first risk level and the second risk level are used to predict the risk level of the target user; wherein, the target group of users is a common user of the enabling institution and the target institution, and the target The private data of the group user includes the private data of the target group user in the enabling organization and the private data of the target group user in the target organization. Optionally, in an embodiment, the third prediction unit 504 is configured to: determine the risk level of the target user based on the first risk level and the second risk level through an additive model. The risk prediction apparatus 500 can implement the methods of the method embodiments shown in FIGS. 1 to 2 . For details, reference may be made to the risk prediction methods of the embodiments shown in FIGS. 1 to 2 , which will not be repeated. FIG. 6 is a schematic structural diagram of a training apparatus 600 for a risk prediction model provided by an embodiment of the present specification. Referring to FIG. 6 , in a software implementation, the training device 600 of the risk prediction model may include: a data acquisition unit 601 , which acquires the private data of the users of the enabling institution and the private data of the users of the target group, wherein the The target group user is a common user of the enabling organization and the target organization, and the private information of the target group user includes the private information of the target group user in the enabling organization, and the target group private data of the user in the target institution, the users of the enabling institution include users of the target group; the first training unit 602, based on the private data of the users of the enabling institution, through isomorphic transfer training obtaining a first risk prediction model; the second training unit 603, based on the private data of the target group users and the corresponding tags, obtains a second risk prediction model through vertical federated learning training, the private data of the target group users corresponds to The label is the fitting error corresponding to the first risk prediction model for the target group users; wherein the first risk prediction model and the second risk prediction model are used to jointly identify the risk level of the user. Optionally, in an implementation manner, the first training unit 602 is configured to: acquire private information of the target group users in the empowerment institution; based on the private information of the users of the empowerment institution and the private data of the target group users in the empowering agency, and obtain the first risk prediction model through isomorphic transfer training. Optionally, in an implementation manner, the second training unit 603 is configured to: obtain the first risk prediction model based on the predicted value of the test data and the actual value corresponding to the test data. Fitting error of the risk prediction model; Based on the private data of the target group users, a second risk prediction model is obtained through longitudinal federated learning training until the predicted value of the second risk prediction model approaches the first risk prediction model fitting error. The risk prediction model training apparatus 600 can implement the methods of the method embodiments shown in FIGS. 3 to 4 . For details, reference may be made to the training methods of the risk prediction model in the embodiments shown in FIGS. 3 to 4 , which will not be repeated. FIG. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present specification. Referring to FIG. 7 , at the hardware level, the electronic device includes a processor, optionally an internal bus, a network interface, and a memory. Wherein, the memory may include internal memory, such as high-speed random-access memory (Random-Access Memory, RAM), and may also include non-volatile memory (non-volatile memory), such as at least one disk memory Wait. Of course, the electronic device may also include hardware required for other services. The processor, the network interface and the memory can be connected to each other through the internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect, peripheral component interconnect standard) bus Or EISA (Extended Industry Standard Architecture, Extended Industry Standard Architecture) bus bar and so on. The busbars can be classified into address busbars, data busbars, control busbars, and the like. For ease of representation, only one double-headed arrow is shown in FIG. 7, but it does not mean that there is only one busbar or one type of busbar. memory for storing programs. Specifically, the program may include program code, and the program code includes computer operation instructions. Memory may include internal memory and non-volatile memory and provide instructions and data to the processor. The processor reads the corresponding computer program from the non-volatile memory into the internal memory and runs it, forming a risk prediction device at the logical level. The processor executes the program stored in the memory, and is specifically configured to perform the following operations: obtain the first private data corresponding to the user ID of the target user from the user database of the enabling organization, and obtain the first private data corresponding to the user ID of the target user from the user database of the enabling organization; Obtaining the second private data corresponding to the user ID of the target user from the user database; inputting the first private data into the first risk level prediction model, and predicting the first private data of the target user the risk level, the first risk level prediction model is obtained through isomorphic transfer training based on the private data of the user of the enabling institution; the first private data and the second private data are input into the second risk level In the prediction model, the second risk level of the target user is predicted, and the second risk level prediction model is obtained through vertical federated learning training based on the private information and corresponding labels of the target group users. The label corresponding to the private data of the group user is the fitting error corresponding to the first risk prediction model of the target group user; based on the first risk level and the second risk level, predict the risk level of the target user ; Wherein, the target group user is a common user of the enabling institution and the target institution, and the private information of the target group user includes the private information of the target group user in the enabling institution data, and private data of users of said target group in the target institution. The above-mentioned methods performed by the risk prediction apparatus disclosed in the embodiments shown in FIG. 1 to FIG. 2 of this specification may be applied to a processor, or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above-mentioned method can be completed through the integrated logic circuit of hardware in the processor or the instructions in the form of software. The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (Digital Signal Processor, DSP), a dedicated Integrated circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Various methods, steps and logic block diagrams disclosed in the embodiments of this specification can be implemented or executed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the methods disclosed in conjunction with the embodiments of the present specification may be directly embodied as being executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in random memory, flash memory, read-only memory, programmable read-only memory or electrically readable and writable programmable memory, temporary storage and other mature storage media in the field. The storage medium is located in the memory, and the processor reads the information in the memory, and completes the steps of the above method in combination with its hardware. The electronic device can also execute the methods shown in FIGS. 1 to 2 , and implement the functions of the risk prediction apparatus in the embodiments shown in FIGS. 1 to 2 , and the embodiments of this specification will not be repeated here. An embodiment of the present specification also provides a computer-readable storage medium, the computer-readable storage medium stores one or more programs, and the one or more programs include instructions, and the instructions, when stored by a portable electronic device including a plurality of application programs When the device is executed, the portable electronic device can be made to execute the method of the embodiment shown in FIG. 1 to FIG. 2 , and is specifically used to perform the following operations: Obtain the usage of the target user from the user database of the empowering organization the first private data corresponding to the user identifier, and the second private data corresponding to the user identifier of the target user obtained from the user database of the target institution; inputting the first private data into the first risk level In the prediction model, the first risk level of the target user is predicted to be obtained, and the first risk level prediction model is obtained through isomorphic migration training based on the private data of the user of the enabling institution; The data and the second private data are input into the second risk level prediction model, and the second risk level of the target user is predicted to be obtained, and the second risk level prediction model is based on the private data of the target group users and the corresponding , obtained through longitudinal federated learning training, the label corresponding to the private data of the target group user is the fitting error corresponding to the target group user in the first risk prediction model; based on the first risk level and the The second risk level, predicting the risk level of the target user; wherein, the target group user is a common user of the enabling organization and the target organization, and the private information of the target group user includes all private data of the target group user in the enabling organization, and private data of the target group user in the target organization. Of course, in addition to software implementations, the electronic devices in this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is to say, the execution subjects of the following processing procedures are not limited to each logic unit , can also be a hardware or logic device. FIG. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present specification. Referring to FIG. 8 , at the hardware level, the electronic device includes a processor, optionally an internal bus, a network interface, and a memory. Wherein, the memory may include internal memory, such as high-speed random-access memory (Random-Access Memory, RAM), and may also include non-volatile memory (non-volatile memory), such as at least one disk memory Wait. Of course, the electronic device may also include hardware required for other services. The processor, the network interface and the memory can be connected to each other through the internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect, peripheral component interconnect standard) bus Or EISA (Extended Industry Standard Architecture, Extended Industry Standard Architecture) bus bar and so on. The busbars can be classified into address busbars, data busbars, control busbars, and the like. For ease of representation, only one double-headed arrow is shown in FIG. 8, but it does not mean that there is only one busbar or one type of busbar. memory for storing programs. Specifically, the program may include program code, and the program code includes computer operation instructions. Memory may include internal memory and non-volatile memory and provide instructions and data to the processor. The processor reads the corresponding computer program from the non-volatile memory into the internal memory and runs it, forming a training device for the risk prediction model at the logical level. The processor executes the program stored in the memory, and is specifically configured to perform the following operations: Acquiring private data of users of the empowerment organization and private data of target group users, wherein the target group users are the users of the empowerment organization The common user of the enabling institution and the target institution, the private information of the target group user includes the private information of the target group user in the enabling institution and the private information of the target group user in the target institution data, the users of the empowerment agency include the target group users; based on the private data of the users of the empowerment agency, a first risk prediction model is obtained through isomorphic transfer training; based on the target group users The private data and corresponding labels are obtained through longitudinal federated learning training to obtain a second risk prediction model, and the label corresponding to the private data of the target group users is the fitting error corresponding to the first risk prediction model of the target group users; wherein , the first risk prediction model and the second risk prediction model are used to jointly identify the user's risk level. The above-mentioned method executed by the apparatus for training the risk prediction model disclosed in the embodiments shown in FIG. 3 to FIG. 4 of this specification may be applied to a processor, or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above-mentioned method can be completed through the integrated logic circuit of hardware in the processor or the instructions in the form of software. The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (Digital Signal Processor, DSP), a dedicated Integrated circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Various methods, steps and logic block diagrams disclosed in the embodiments of this specification can be implemented or executed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the methods disclosed in conjunction with the embodiments of the present specification may be directly embodied as being executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in random memory, flash memory, read-only memory, programmable read-only memory or electrically readable and writable programmable memory, temporary storage and other mature storage media in the field. The storage medium is located in the memory, and the processor reads the information in the memory, and completes the steps of the above method in combination with its hardware. The electronic device can also execute the methods shown in FIGS. 3 to 4 , and implement the functions of the training device for the risk prediction model in the embodiments shown in FIGS. An embodiment of the present specification also provides a computer-readable storage medium, the computer-readable storage medium stores one or more programs, and the one or more programs include instructions, and the instructions, when stored by a portable electronic device including a plurality of application programs When the device is executed, the portable electronic device can be made to execute the methods of the embodiments shown in FIG. 3 to FIG. 4 , and is specifically used to perform the following operations: Obtaining the private information of the users of the enabling organization and the private information of the users of the target group data, wherein the target group user is a common user of the enabling organization and the target organization, and the private data of the target group user includes the private data of the target group user in the enabling organization , and the private data of the target group users in the target organization, the users of the enabling organization include the target group users; based on the private data of the users of the enabling organization, through isomorphic transfer training Obtain a first risk prediction model; Based on the private data of the target group users and corresponding labels, obtain a second risk prediction model through longitudinal federated learning training, and the labels corresponding to the private data of the target group users are used by the target group The fitting error corresponding to the user in the first risk prediction model; wherein, the first risk prediction model and the second risk prediction model are used to jointly identify the risk level of the user. The foregoing describes specific embodiments of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. Additionally, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multiplexing and parallel processing are also possible or may be advantageous. In a word, the above descriptions are only preferred embodiments of the present specification, and are not intended to limit the protection scope of the present specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification shall be included within the protection scope of this specification. The systems, devices, modules or units described in the above embodiments may be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device Or a combination of any of these devices. Computer-readable media includes both permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology for storage of information. Information can be computer readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change internal memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM) ), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other internal memory technology, compact disk read only memory (CD-ROM), digital A Versatile Disc (DVD) or other optical storage, magnetic tape cassette, magnetic tape storage or other magnetic storage device or any other non-transmission medium may be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transient media, such as modulated data signals and carrier waves. It should also be noted that the terms "comprising", "comprising" or any other variation thereof are intended to encompass a non-exclusive inclusion such that a process, method, article or device comprising a series of elements includes not only those elements, but also Other elements not expressly listed, or inherent to such a process, method, article of manufacture or apparatus are also included. Without further limitation, an element qualified by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article of manufacture, or device that includes the element. Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for related parts, please refer to the partial descriptions of the method embodiments.