CN113807530B - Information processing system, method and device - Google Patents
Information processing system, method and device Download PDFInfo
- Publication number
- CN113807530B CN113807530B CN202011018697.7A CN202011018697A CN113807530B CN 113807530 B CN113807530 B CN 113807530B CN 202011018697 A CN202011018697 A CN 202011018697A CN 113807530 B CN113807530 B CN 113807530B
- Authority
- CN
- China
- Prior art keywords
- encryption
- processed
- leaf node
- leaf nodes
- tree model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000010365 information processing Effects 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000004044 response Effects 0.000 claims abstract description 77
- 238000012545 processing Methods 0.000 claims abstract description 72
- 238000012216 screening Methods 0.000 claims abstract description 53
- 238000003672 processing method Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 238000012549 training Methods 0.000 description 10
- 238000004891 communication Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioethics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Databases & Information Systems (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Mathematical Physics (AREA)
Abstract
The embodiment of the disclosure discloses an information processing system, method and device. One embodiment of the system comprises: a first end configured to: screening out target leaf nodes which cannot be reached by the object to be processed based on the first characteristic information of the object to be processed; for a leaf node included in the tree model, encrypting 0 in response to the leaf node being a target leaf node to obtain an encryption weight; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain an encrypted weight; sending encryption weight and a processing request aiming at an object to be processed to a second end; a second end configured to: screening out at least two candidate leaf nodes which are possibly reached by the object to be processed based on the second characteristic information of the object to be processed; summing the encryption weights of at least two candidate leaf nodes to obtain an encryption score; the encrypted result is sent to the first end. The embodiment can improve the security of the first end data.
Description
Technical Field
Embodiments of the present disclosure relate to the field of computer technology, and more particularly, to information processing systems, methods, and apparatuses.
Background
In machine learning, the data determines an upper limit of the model effect, and in order to further improve the accuracy of the model, the business side can aggregate the data of the data side to perform federal learning of the model. The business party can be a party which provides feature data and a label of the feature data simultaneously in the process of training the model, and the data party can be a party which only provides the feature data.
In practice, the model obtained by training can be deployed on the business side and the data side respectively, but the business side and the data side only manage partial characteristics and characteristic thresholds of the model.
Because the service party only manages part of the characteristics and the characteristic threshold values of the model, the service party also needs to cooperate with the data party to complete the prediction process when the model is utilized for prediction.
Disclosure of Invention
The embodiment of the disclosure provides an information processing system, method and device.
In a first aspect, embodiments of the present disclosure provide an information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing partial nodes in the tree model, respectively, the first end having weights for leaf nodes of the tree model, wherein the first end is configured to: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; sending the obtained encryption weight and a processing request for the object to be processed to a second end; the second end is configured to: acquiring second characteristic information of the object to be processed in response to receiving the encryption weight sent by the first end and the processing request aiming at the object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model; the encrypted result is sent to the first end.
In some embodiments, the tree model includes at least two trees; and the second end is further configured to: for a tree of at least two trees comprised by the tree model, the following steps are performed: screening at least two leaf nodes which are possibly reached by an object to be processed from leaf nodes included in the tree as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree; and summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model.
In some embodiments, the first end is further configured to: decrypting the received encrypted score to obtain a target score; the object to be processed is classified based on the target score.
In some embodiments, the first end is further configured to: in response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node; and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node. In a second aspect, an embodiment of the present disclosure provides an information processing method performed by a first end in an information processing system including a first end and a second end, the first end and the second end having the same tree model, the first end and the second end managing part of nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the method comprising: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; sending the obtained encryption weight and a processing request for the object to be processed to a second end; and receiving an encryption score from the second end, wherein the target encryption score is obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
In some embodiments, the method further comprises: decrypting the received encrypted score to obtain a target score; the object to be processed is classified based on the target score.
In a third aspect, an embodiment of the present disclosure provides an information processing method performed by a second terminal in an information processing system, the information processing system including a first terminal and a second terminal, the first terminal and the second terminal having the same tree model, the first terminal and the second terminal managing part of nodes in the tree model, respectively, the first terminal having weights of leaf nodes of the tree model, the method comprising: in response to receiving the encryption weight sent by the first end and the processing request aiming at the object to be processed, obtaining second characteristic information of the object to be processed, wherein the encryption weight is obtained by the first end through the following steps: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model; the encrypted result is sent to the first end.
In some embodiments, the tree model includes at least two trees; screening at least two leaf nodes which are possibly reached by the object to be processed from the leaf nodes included in the tree model based on the second characteristic information, and taking the leaf nodes as at least two candidate leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes, wherein obtaining the encryption score of the object to be processed in the tree model comprises the following steps: for a tree of at least two trees comprised by the tree model, the following steps are performed: screening at least two leaf nodes which are possibly reached by an object to be processed from leaf nodes included in the tree as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree; and summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model.
In a fourth aspect, an embodiment of the present disclosure provides an information processing apparatus configured at a first end in an information processing system, the information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing part of nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the apparatus comprising: a first acquisition unit configured to acquire first characteristic information of an object to be processed; a first screening unit configured to screen, as a target leaf node, a leaf node from among leaf nodes included in the tree model that a to-be-processed object does not reach, based on the first feature information; an execution unit configured to execute, for leaf nodes included in the tree model, the following processing steps: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; a first transmitting unit configured to transmit the obtained encryption weight and a processing request for an object to be processed to a second end; and a receiving unit configured to receive the encryption score from the second end, wherein the target encryption score is obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
In some embodiments, the apparatus further comprises: a decryption unit configured to decrypt the received encryption score to obtain a target score; and a classification unit configured to classify the object to be processed based on the target score.
In some embodiments, the execution unit is further configured to: in response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node; and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node. In a fifth aspect, an embodiment of the present disclosure provides an information processing apparatus configured at a second end in an information processing system, the information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing part of nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the apparatus comprising: a second obtaining unit configured to obtain second feature information of the object to be processed in response to receiving the encryption weight transmitted by the first end and the processing request for the object to be processed, wherein the encryption weight is obtained by the first end by: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; a second screening unit configured to screen at least two leaf nodes, which are possible to reach the object to be processed, from leaf nodes included in the tree model as at least two candidate leaf nodes based on the second feature information, wherein the at least two candidate leaf nodes include target leaf nodes; the summing unit is configured to sum the encryption weights corresponding to the at least two candidate leaf nodes to obtain encryption scores of the objects to be processed in the tree model; and a second transmitting unit configured to transmit the encrypted result to the first terminal.
In some embodiments, the tree model includes at least two trees; and the second screening unit and the summing unit are further configured to: for a tree of at least two trees comprised by the tree model, the following steps are performed: screening at least two leaf nodes which are possibly reached by an object to be processed from leaf nodes included in the tree as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree; and summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model.
In a sixth aspect, embodiments of the present disclosure provide an electronic device, comprising: one or more processors; and a storage device having one or more programs stored thereon, which when executed by the one or more processors, cause the one or more processors to implement the method of any of the embodiments of the information processing method described above.
In a seventh aspect, embodiments of the present disclosure provide a computer-readable medium having stored thereon a computer program which, when executed by a processor, implements a method of any of the embodiments of the information processing method described above.
In practice, the interpretability of models is of great importance in many fields. The interpretable model may provide a basis for it to obtain a certain decision. Currently, in the federal learning field, in order to improve the interpretability of a model, a business party can disclose to a data party the characteristics corresponding to the nodes it manages. However, in the process of predicting by using the model, the service side will also typically disclose to the data side the splitting direction of the object to be predicted at the node managed by the service side, so the data side can obtain the feature corresponding to the node managed by the service side and the splitting direction of the object to be predicted under the feature at the same time, which may enable the data side to infer the feature value corresponding to the node managed by the service side based on the splitting direction and the feature, in combination with background knowledge or elimination method, and further cause privacy leakage of the service side.
According to the information processing system, the information processing method and the information processing device, first characteristic information of an object to be processed can be obtained through the first end, then leaf nodes which cannot be reached by the object to be processed are screened out of leaf nodes included in the tree model to serve as target leaf nodes based on the first characteristic information, and then the following processing steps are executed for the leaf nodes included in the tree model: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being a target leaf node, encrypting the weight of the leaf node to obtain an encryption weight corresponding to the leaf node, then sending the obtained encryption weight and a processing request aiming at an object to be processed to a second end, then the second end can obtain second characteristic information of the object to be processed in response to receiving the encryption weight sent by the first end and the processing request aiming at the object to be processed, and then based on the second characteristic information, at least two leaf nodes which possibly reach the object to be processed are screened out of the leaf nodes included in a tree model to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes, then the encryption weights corresponding to the at least two candidate leaf nodes are summed to obtain an encryption score of the object to be processed in the tree model, and finally the encryption score is sent to the first end, so that the first end can respectively encrypt virtual weights (namely 0) of the target leaf nodes and real weights of the leaf nodes except the target leaf nodes and then send the second end, the second end cannot be distinguished by the encryption technology, the second end can not know the target node, the privacy of the object to be possibly reached by the leaf nodes can be further, the first end can not be accurately obtained, the first end can further obtain the security score of the object to be better the first end, and the security of the object to be better is better than the first end can be compared, and the security can be better by the security of the first end; in addition, the second end in the present disclosure only needs to send the encrypted result obtained by summation to the first end, and does not need to disclose the splitting direction to the first end, so the present disclosure can also protect the privacy of the second end.
Drawings
Other features, objects and advantages of the present disclosure will become more apparent upon reading of the detailed description of non-limiting embodiments, made with reference to the following drawings:
FIG. 1 is an exemplary system architecture diagram of an information handling system of the present disclosure;
FIG. 2 is a timing diagram of one embodiment of an information handling system according to the present disclosure;
FIG. 3 is a schematic diagram of the structure of a tree model in one embodiment of a system for processing information according to the present disclosure;
FIG. 4 is a flow chart of one embodiment of an information processing method according to the present disclosure;
FIG. 5 is a flow chart of yet another embodiment of an information processing method according to the present disclosure;
FIG. 6 is a schematic structural view of one embodiment of an information processing apparatus according to the present disclosure;
fig. 7 is a schematic structural view of still another embodiment of an information processing apparatus according to the present disclosure;
fig. 8 is a schematic diagram of a computer system suitable for use in implementing embodiments of the present disclosure.
Detailed Description
The present disclosure is described in further detail below with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings.
It should be noted that, without conflict, the embodiments of the present disclosure and features of the embodiments may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
FIG. 1 illustrates an exemplary system architecture 100 of an information handling system of the present disclosure.
As shown in fig. 1, the system architecture 100 may include first ends 101, 102, a network 103, and second ends 104, 105. The network 103 is used as a medium to provide a communication link between the first ends 101, 102 and the second ends 104, 105. The network 103 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
The first end 101, 102 and the second end 104, 105 may be participating ends of federal learning of the model. Specifically, the first ends 101, 102 may correspond to business parties in the federal learning process, and the second ends 104, 105 may correspond to data parties in the federal learning process. Wherein the business party may be a party who provides feature data and a tag of the feature data simultaneously in the federal learning process. The data party may be a party that only provides the feature data. The dimensions of the feature data provided by the business party are typically different from the dimensions of the data provided by the data party (e.g., the business party provides feature data for an age feature, and the data party provides feature data for a height feature). The federally learned acquisition models may be deployed in the first and second ends 101, 102, 104, 105, respectively. The first ends 101, 102 and the second ends 104, 105 respectively manage features corresponding to the feature data provided by them.
The first end 101, 102 may be hardware or software. When the first end 101, 102 is hardware, it may be a variety of electronic devices including, but not limited to, servers, smartphones, tablets, laptop and desktop computers, and the like. When the first end 101, 102 is software, it may be installed in the electronic device listed above. Which may be implemented as multiple software or software modules (e.g., multiple software or software modules for providing distributed services) or as a single software or software module. The present invention is not particularly limited herein.
Similarly, the second ends 104, 105 may be hardware or software. When the second end 104, 105 is hardware, it may be a variety of electronic devices including, but not limited to, servers, smartphones, tablets, laptop and desktop computers, and the like. When the second end 104, 105 is software, it may be installed in the electronic device as listed above. Which may be implemented as multiple software or software modules (e.g., multiple software or software modules for providing distributed services) or as a single software or software module. The present invention is not particularly limited herein.
It should be understood that the number of first ends, networks, and second ends in fig. 1 are merely illustrative. There may be any number of first ends, networks, and second ends, as desired for implementation.
Referring to FIG. 2, a timing diagram 200 of one embodiment of an information handling system according to the present application is shown.
The information processing system in the embodiment of the application may include a first end and a second end, where the first end and the second end may have the same tree model, and the tree model may be various models composed of trees, for example, may be a gradient lifting tree model, a random forest, and the like. In particular, the tree model may be a model obtained by joint learning training of the first end and the second end. The first and second ends may each manage a portion of the nodes in the tree model.
Specifically, as an example, please refer to fig. 3, fig. 3 shows a schematic structural diagram of a tree model according to the present embodiment. The tree model may be configured at a first end and a second end, respectively. In particular, the tree model may include three nodes 301, 302, and 303, and four leaf nodes 304, 305, 306, and 307. Node 301 of the three nodes may be a node managed by a first end and nodes 302 and 303 may be nodes managed by a second end.
In this embodiment, the nodes in the tree model managed by the first end and the second end respectively may be determined in the process of training to obtain the tree model. Specifically, the nodes managed by the first end and the second end are related to the feature information that can be acquired by the first end and the second end. As an example, if the first end can obtain the age information and the second end can obtain the height information, the first end manages the node corresponding to the age feature in the tree model, and the second end manages the node corresponding to the height feature in the tree model.
In this embodiment, the first end may have the weight of the leaf nodes of the tree model. Here, the nodes of the tree model may be split, while the leaf nodes of the tree model may not be split. The weights of the leaf nodes may be obtained during training of the tree model. Specifically, the weight of a leaf node may be used to indicate the number of training objects in the training object set that fall on the leaf node. The greater the number, the greater the weight of the leaf node may be.
It should be noted that, the method of obtaining the tree model through joint learning training is a well-known technique widely studied and applied at present, and will not be described herein.
In this embodiment, the first end may be configured to: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; and sending the obtained encryption weight and a processing request for the object to be processed to a second end. The second end may be configured to: acquiring second characteristic information of the object to be processed in response to receiving the encryption weight sent by the first end and the processing request aiming at the object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model; the encrypted result is sent to the first end.
Specifically, as shown in fig. 2, in step 201, a first end acquires first feature information of an object to be processed.
In this embodiment, the first end (for example, the first end 101 shown in fig. 1) may acquire the first feature information of the object to be processed from a remote location or a local location through a wired connection manner or a wireless connection manner. The object to be processed may be an object to be processed, specifically may be various objects that can be processed by the tree model, for example, may be an image, text, video, audio, and the like. The first characteristic information of the object to be processed may be used to characterize the characteristics of the object to be processed. Specifically, the characteristic represented by the first characteristic information acquired by the first end may be the same as the characteristic corresponding to the node managed by the first end.
As an example, the feature corresponding to the node managed by the first end is an age feature, and the first feature information may be information for characterizing the age feature of the object to be processed (e.g., user).
In step 202, the first end screens out leaf nodes, which the object to be processed will not reach, from the leaf nodes included in the tree model as target leaf nodes based on the first feature information.
In this embodiment, based on the first feature information obtained in step 201, the first end may split the object to be processed in the tree model, so as to screen out leaf nodes that the object to be processed will not reach from leaf nodes included in the tree model as target leaf nodes.
Specifically, the first end may determine, based on the first feature information, a leaf node and a target leaf node that may be reached by the object to be processed from leaf nodes included in the tree model, where the target leaf node is a leaf node in the tree model other than the leaf node that may be reached by the object to be processed.
It will be appreciated that there are nodes managed by the first end and nodes managed by the second end in the tree model. For a node managed by a first end in the tree model, the first end can determine a splitting direction of an object to be processed in the node based on the first characteristic information, and then can determine a target leaf node which the object to be processed cannot reach according to the splitting direction. In particular, the first end may also determine, according to the splitting direction, a leaf node that the object to be processed can reach. For the node managed by the second end in the tree model, the first end cannot know the splitting direction of the object to be processed in the node, in this case, the first end may use each splitting direction of the object to be processed in the node as a possible splitting direction, and further use the leaf node corresponding to the node as a leaf node where the object to be processed may arrive. Thus, the first end can determine the leaf nodes which the object to be processed possibly reaches and the target leaf nodes which the object to be processed cannot reach from the tree model based on the first characteristic information.
It should be noted that, in the present disclosure, it is not possible to determine, based on the feature information (the first feature information or the second feature information), a leaf node that the object to be processed can reach (specifically, determined by the structure of the tree model and the nodes managed by the first end and the second end respectively), and if it is determined, based on the feature information, that the leaf node that the object to be processed can reach is also referred to as a leaf node that the object to be processed can reach for convenience of expression.
Continuing with the example of fig. 3, the node 301 in the tree model of fig. 3 is a node managed by the first end, and then the first end may determine, based on the first feature information, a splitting direction of the object to be processed at the node 301, for example, splitting to the right, then the object to be processed may reach the node 303 after the first splitting, and the node 302 is a node that the object to be processed does not reach, and further, the two leaf nodes 304, 305 connected to the node 302 are leaf nodes that the object to be processed does not reach, that is, target leaf nodes. After the object to be processed reaches the node 303, a second splitting is required, however, since the node 303 is a node managed by the second end, the first end cannot learn the splitting direction of the object to be processed at the node 303, and from the perspective of the first end, the object to be processed may split left or split right at the node 303, so the first end may use two leaf nodes 306 and 306 connected to the node 303 as leaf nodes that the object to be processed may reach.
In step 203, the first end performs a processing step for the leaf nodes comprised by the tree model.
In this embodiment, for each of the leaf nodes included in the tree model, the first end may perform the following processing steps (steps 2031-2032):
in step 2031, in response to the leaf node being the target leaf node, encryption is performed on 0 to obtain an encryption weight corresponding to the target leaf node.
In step 2032, in response to the leaf node not being the target leaf node, the weight of the leaf node is encrypted, and the encryption weight corresponding to the leaf node is obtained.
Specifically, the first end may encrypt the weight of the 0 or leaf node in a homomorphic encryption manner, so that an encryption manner corresponding to an encryption score obtained by summing the encryption weights in the following process is the same as an encryption manner corresponding to the encryption weight, and further the first end may decrypt the encryption score based on the encryption manner of encrypting the weight of the 0 or leaf node after obtaining the encryption score, to obtain the score of the object to be processed.
In practice, in order to determine the leaf node actually reached by the object to be processed in the tree model, the second end needs to know the leaf node possibly reached by the object to be processed at the first end, and then the intersection is taken between the leaf node possibly reached by the object to be processed at the first end and the leaf node possibly reached by the object to be processed at the second end, so that the leaf node actually reached by the object to be processed in the tree model can be obtained. However, if the first end directly discloses the leaf node that the object to be processed may reach at the first end to the second end, the second end will know the target leaf node that the object to be processed will not reach at the first end, which will enable the second end to know the splitting direction of the object to be processed at the node managed by the first end, thereby causing privacy leakage at the first end. For example, continuing with the example of fig. 3, if the first end discloses to the second end that it is possible to reach the leaf node 306 and the leaf node 307, the second end may learn that the object to be processed does not reach the leaf node 305 and the leaf node 306, and may further learn that the splitting direction of the object to be processed at the node 301 managed by the first end is split rightward.
The first end in the solution of this embodiment may disclose the weight of the leaf node to the second end in an encrypted manner, which will make it impossible for the second end to distinguish, according to the encrypted weight, a leaf node that the object to be processed may reach at the first end from a target leaf node that will not reach. And for the target leaf node which the object to be processed cannot reach at the first end, 0 can be used as the virtual weight of the target leaf node, the encryption weight obtained by encrypting 0 is retransmitted to the second end, and then when the score of the object to be processed is obtained by summation, the target leaf node can not influence the summation result, so that the accuracy of the obtained score of the object to be processed can be ensured.
Continuing with the example of fig. 3, for four leaf nodes in the tree model shown in fig. 3, the leaf nodes 304, 305 are target leaf nodes that the object to be processed will not reach, the first end may encrypt 0 with 0 as the virtual weights of the two target leaf nodes, and obtain the encryption weights x corresponding to the target leaf nodes 304, 305; the leaf nodes 306 and 307 are possible leaf nodes that the object to be processed may reach, the first end may obtain the weight "0.4" of the leaf node 306 and obtain the weight "0.2" of the leaf node 307, then encrypt the weight "0.4" to obtain the encryption weight y corresponding to the leaf node 306, and encrypt the weight "0.2" to obtain the encryption weight z corresponding to the leaf node 307.
In step 204, the first end sends the obtained encryption weight and a processing request for the object to be processed to the second end.
In this embodiment, based on the encryption weight obtained in step 203, the first end may retransmit the encryption weight to the second end (for example, the second end shown in fig. 1), and send a processing request for the above-described object to be processed to the second end. Wherein the processing request may be used to request processing of the object to be processed. Specifically, the processing request may include an identifier (e.g., an ID) of the object to be processed, and after the second end receives the processing request, the object to be processed may be processed based on the identifier of the object to be processed.
Continuing with the example above, the first end may send the encryption weights x, y, z described above to the second end. Furthermore, the second end can only obtain the encryption information, and cannot directly or indirectly obtain the splitting direction of the object to be processed at the first end.
In step 205, the second end obtains second feature information of the object to be processed in response to receiving the encryption weight sent by the first end and the processing request for the object to be processed.
In this embodiment, the second end may obtain, in response to receiving the encryption weight and the processing request sent by the first end, second feature information of the object to be processed from a remote location or a local location through a wired connection manner or a wireless connection manner. Wherein the second characteristic information of the object to be processed may be used for characterizing the characteristics of the object to be processed. Specifically, the characteristic represented by the second characteristic information acquired by the second end may be the same as the characteristic corresponding to the node managed by the second end.
As an example, the feature corresponding to the node managed by the second end is a height feature, and the second feature information may be information for characterizing the height feature of the object to be processed (e.g., the user).
In step 206, the second end screens out at least two leaf nodes that the object to be processed may reach from the leaf nodes included in the tree model as at least two candidate leaf nodes based on the second feature information.
In this embodiment, based on the second feature information obtained in step 205, the second end may screen at least two leaf nodes that may reach the object to be processed from the leaf nodes included in the tree model as at least two candidate leaf nodes.
Specifically, for a node managed by the second end in the tree model, the second end can determine a splitting direction of the object to be processed at the node based on the second characteristic information, and further determine a leaf node that the object to be processed cannot reach after splitting at the node according to the splitting direction. In particular, the second end may also determine, according to the splitting direction, a leaf node that the object to be processed can reach. For the node managed by the first end in the tree model, the second end cannot know the splitting direction of the object to be processed in the node, in this case, the second end may take each splitting direction of the object to be processed in the node as a possible splitting direction, and further take leaf nodes corresponding to the node as leaf nodes where the object to be processed may possibly arrive. In this way, the second end may determine, from the tree model, at least two candidate leaf nodes that the object to be processed may reach based on the second feature information.
It will be appreciated that, here, since leaf nodes corresponding to the node managed by the first end are all candidate leaf nodes, and leaf nodes under the node managed by the first end necessarily include leaf nodes that the object to be processed may reach at the first end and target leaf nodes that will not reach, the candidate leaf nodes include leaf nodes that the object to be processed may reach at the first end and target leaf nodes that will not reach.
Continuing with the above example, after the second end obtains the encryption weights x, y, z, the object to be processed may be split in the tree model based on the second feature information. Specifically, in the corresponding node 301, since the node 301 is a node managed by the first end, the second end cannot learn the splitting direction of the object to be processed in the node 301, and from the perspective of the second end, the object to be processed may split left or split right in the node 301, so that the second end may use both the node 302 and the node 303 as nodes where the object to be processed may reach. Then for node 302, the second end may determine, based on the second characteristic information, a splitting direction of the object to be processed at node 302, e.g. splitting to the right, and the second end may determine that leaf node 305 is a candidate leaf node where the object to be processed may arrive at the second end. For node 303, the second end may determine, based on the second characteristic information, a splitting direction of the object to be processed at node 303, e.g., to the left, and the second end may determine that leaf node 306 is a candidate leaf node that the object to be processed may reach at the second end. Further, the second end may obtain two candidate leaf nodes, candidate leaf node 305 and candidate leaf node 306, respectively, based on the second feature information. Therein, as can be seen from the above example, the candidate leaf node 305 is a target leaf node that the object to be processed will not reach at the first end.
Further, in the above example, the candidate leaf node 306 is an intersection between a leaf node that the object to be processed may reach at the first end and a leaf node that the object to be processed may reach at the second end, and further, it is known that the candidate leaf node 306 is a leaf node that the object to be processed actually reaches in the tree model.
In step 207, the second end sums the encryption weights corresponding to the at least two candidate leaf nodes to obtain an encryption score of the object to be processed in the tree model.
In this embodiment, based on the at least two candidate leaf nodes screened in step 206, the second end may extract at least two encryption weights corresponding to the screened at least two candidate leaf nodes from the received encryption weights, and sum the extracted at least two encryption weights to obtain the encryption score of the object to be processed in the tree model. Wherein the encryption score may be a score of the encrypted object to be processed. The score of the object to be processed may be a numerical value for characterizing the likelihood that the object to be processed belongs to a preset class. The higher the score, the greater the likelihood that the object to be processed belongs to the preset category may be characterized. The preset category may be a category that the tree model can predict.
Continuing with the above example, for the screened candidate leaf node 305 and candidate leaf node 306, the second end may obtain the encryption weight x corresponding to the candidate leaf node 305 and the encryption weight y corresponding to the candidate leaf node, and sum the two to obtain the encryption score "x+y".
In practice, the number of trees included in the tree model may be arbitrary, and in some alternative implementations of the present embodiment, the tree model may include at least two trees; and the second end may perform the above-described step 206 and the above-described step 207 by:
first, for a tree of at least two trees comprised by the tree model, the second end may perform the steps of: screening at least two leaf nodes which are possibly reached by an object to be processed from leaf nodes included in the tree as at least two candidate leaf nodes; and summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree.
The second end may then sum at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain an encryption score for the object to be processed in the tree model.
The tree model in the implementation manner comprises at least two trees, and further when the second end generates encryption scores, at least two initial encryption scores corresponding to the at least two trees respectively need to be summed, so that the encryption scores sent to the first end by the second end are obtained through statistics on the results of the trees, when the first end cannot obtain the splitting results corresponding to the trees in the tree model for splitting the object to be processed, the possibility that the first end presumes the splitting direction of the object to be processed in the second end based on the splitting results of the trees can be reduced, privacy of the second end is protected, and safety of data of the second end is improved.
In step 208, the second end sends the encrypted result to the first end.
In this embodiment, after obtaining the encryption score, the second end may send the encryption score to the first end, so that the first end classifies the object to be processed based on the encryption score.
In some optional implementations of this embodiment, after the first end receives the encryption score, the following steps may be performed:
and a first step of decrypting the received encrypted score to obtain a target score.
Specifically, the first end may decrypt the encrypted score based on the encryption of the weight of the 0 or leaf node in step 203, to obtain the target score.
Continuing with the example above, the first end may receive the encryption score "x+y" sent by the second end and decrypt "x+y" to obtain a target score of 0.4.
It will be appreciated that, since the encryption weight "x" corresponding to the leaf node 305 is the encryption of the virtual weight "0", the decryption result obtained when decrypting the "x" is "0", and thus the decryption result corresponding to the "x" does not affect the target score obtained by summing, so that the leaf node 305 where the object to be processed does not reach at the first end can be removed from the leaf nodes for generating the target score, and the target score determined by the weight corresponding to the leaf node 306 where the object to be processed actually reaches in the tree model can be obtained. With this, the validity and accuracy of the obtained target score can be improved.
And secondly, classifying the object to be processed based on the target score.
Based on the target score obtained in the first step, the first end may classify the object to be processed.
Specifically, as an example, the first end may determine whether the target score is greater than or equal to a target threshold, and in response to the target score being greater than or equal to the target threshold, classify the object to be processed (e.g., the user) as a first class (e.g., an obese class); in response to the target score being less than the target threshold, the object to be processed is classified into a second class (e.g., a non-obese class).
The first end in the system provided by the embodiment of the present disclosure may encrypt the virtual weight (i.e. 0) of the target leaf node and the real weight of the leaf nodes other than the target leaf node, and then send the encrypted virtual weight and the real weight of the leaf nodes other than the target leaf node to the second end, where the encryption technology may enable the second end to fail to distinguish the target leaf node, and further fail to know the splitting direction of the first end, and encryption of 0 may enable the contribution degree of the score of the target leaf node to the object to be processed to be 0, and further may ensure the accuracy of the score of the object to be processed obtained by summation, so the present disclosure may protect the privacy of the first end and improve the security of the data of the first end under the condition that the first end can obtain the accurate score of the object to be processed; in addition, the second end in the present disclosure only needs to send the encrypted result obtained by summation to the first end, and does not need to disclose the splitting direction to the first end, so the present disclosure can also protect the privacy of the second end.
With continued reference to fig. 4, a flow 400 of one embodiment of an information processing method according to the present disclosure is shown. The information processing method comprises the following steps:
step 401, obtaining first feature information of an object to be processed.
In this embodiment, the first end of the execution body of the information processing method (for example, the first end shown in fig. 1) may acquire the first feature information of the object to be processed from a remote location or a local location by a wired connection manner or a wireless connection manner. Wherein the first end belongs to a predetermined information processing system (e.g., the corresponding information processing system of fig. 2). The information handling system may include a first end and a second end, and the first end and the second end may have the same tree model, which may be various models composed of trees. In particular, the tree model may be a model obtained by joint learning training of the first end and the second end. The first and second ends may each manage a portion of the nodes in the tree model. The first end may have weights for leaf nodes of the tree model. Here, the nodes of the tree model may be split, while the leaf nodes of the tree model may not be split.
In this embodiment, the object to be processed may be an object to be processed, specifically may be various objects that can be processed by the above tree model, and may be, for example, an image, text, video, audio, and the like. The first characteristic information of the object to be processed may be used to characterize the characteristics of the object to be processed. Specifically, the characteristic represented by the first characteristic information acquired by the first end may be the same as the characteristic corresponding to the node managed by the first end.
Step 402, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes based on the first characteristic information.
In this embodiment, based on the first feature information obtained in step 401, the first end may screen, as the target leaf node, a leaf node that the object to be processed will not reach from leaf nodes included in the tree model.
Step 403, for the leaf nodes comprised by the tree model, executing the processing steps.
In this embodiment, for the leaf nodes included in the tree model, the first end may perform the following processing steps (step 4031-step 4032):
and step 4031, in response to the leaf node being the target leaf node, encrypting 0 to obtain the encryption weight corresponding to the target leaf node.
In step 4032, in response to the leaf node not being the target leaf node, the weight of the leaf node is encrypted to obtain the encryption weight corresponding to the leaf node.
In some alternative implementations of this embodiment, the first end may perform steps 4031 and 4032 described above in the following manner:
in response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node; and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node. And step 404, sending the obtained encryption weight and a processing request for the object to be processed to a second end.
In this embodiment, based on the encryption weight obtained in step 403, the first end may retransmit the encryption weight to the second end (for example, the second end shown in fig. 1), and send a processing request for the above-described object to be processed to the second end. Wherein the processing request may be used to request processing of the object to be processed.
Step 405 receives an encryption score from a second end.
In this embodiment, after transmitting the encryption weight and the processing request to the second end, the first end may receive the encryption score generated based on the encryption weight from the second end. Wherein the encryption score may be a score of the encrypted object to be processed. The score of the object to be processed may be a numerical value for characterizing the likelihood that the object to be processed belongs to a preset class. The higher the score, the greater the likelihood that the object to be processed belongs to the preset category may be characterized. The preset category may be a category that the tree model can predict.
Specifically, the encryption score may be obtained by the second end by:
the first step, second characteristic information of the object to be processed is obtained.
Wherein the second characteristic information of the object to be processed may be used for characterizing the characteristics of the object to be processed. Specifically, the characteristic represented by the second characteristic information acquired by the second end may be the same as the characteristic corresponding to the node managed by the second end.
And a second step of screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes.
Wherein the at least two candidate leaf nodes comprise a target leaf node.
And thirdly, summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
In some optional implementations of the present embodiment, after receiving the encryption score, the first end may further perform the following steps:
and a first step of decrypting the received encrypted score to obtain a target score.
And secondly, classifying the object to be processed based on the target score.
It should be noted that the first end and the second end in the embodiment corresponding to fig. 4 may have similar or identical features to the first end and the second end in the embodiment corresponding to fig. 2, and the descriptions of the first end and the second end in the embodiment corresponding to fig. 2 also apply to the embodiment corresponding to fig. 4. In order to avoid repetition, the description is omitted here.
The method provided by the embodiment of the present disclosure may be executed by a first end in an information processing system, where the first end may encrypt a virtual weight (i.e. 0) of a target leaf node and a real weight of a leaf node other than the target leaf node, and then send the encrypted virtual weight and the real weight of the leaf node other than the target leaf node to a second end, where the second end cannot distinguish the target leaf node, and thus cannot learn a splitting direction of the first end, and encryption of 0 may further enable the contribution degree of the target leaf node to a score of an object to be processed to be 0, and thus may ensure accuracy of the score of the object to be processed obtained by summation.
With further reference to fig. 5, a flow 500 of yet another embodiment of an information processing method is shown. The flow 500 of the information processing method includes the steps of:
in step 501, second feature information of the object to be processed is obtained in response to receiving the encryption weight sent by the first end and the processing request for the object to be processed.
In this embodiment, the second end of the execution body of the information processing method may acquire the second feature information of the object to be processed in response to receiving, by a wired connection manner or a wireless connection manner, the encryption weight sent by the first end and the processing request for the object to be processed. Wherein the second end belongs to a predetermined information processing system (e.g., the corresponding information processing system of fig. 2). The information handling system may include a first end and a second end, and the first end and the second end may have the same tree model, which may be various models composed of trees. In particular, the tree model may be a model obtained by joint learning training of the first end and the second end. The first and second ends may each manage a portion of the nodes in the tree model. The first end may have weights for leaf nodes of the tree model. Here, the nodes of the tree model may be split, while the leaf nodes of the tree model may not be split.
In this embodiment, the obtained encryption weights may include encryption weights corresponding to respective leaf nodes in the tree model. The second characteristic information of the object to be processed may be used to characterize the characteristics of the object to be processed. Specifically, the characteristic represented by the second characteristic information acquired by the second end may be the same as the characteristic corresponding to the node managed by the second end. The processing request may be used to request processing of the object to be processed.
In this embodiment, the encryption weight may be obtained by the first end by:
first, first characteristic information of an object to be processed is obtained.
Wherein the first characteristic information of the object to be processed may be used for characterizing the characteristics of the object to be processed. Specifically, the characteristic represented by the first characteristic information acquired by the first end may be the same as the characteristic corresponding to the node managed by the first end.
And a second step of screening leaf nodes which are not reached by the object to be processed from the leaf nodes included in the tree model based on the first characteristic information to serve as target leaf nodes.
Third, for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node.
Step 502, based on the second characteristic information, at least two leaf nodes, which are possibly reached by the object to be processed, are selected from leaf nodes included in the tree model as at least two candidate leaf nodes.
In this embodiment, based on the second feature information obtained in step 501, the second end may screen at least two leaf nodes that may reach the object to be processed from the leaf nodes included in the tree model as at least two candidate leaf nodes. Wherein the at least two candidate leaf nodes may include a target leaf node screened out by the first end.
And step 503, summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
In this embodiment, based on the at least two candidate leaf nodes screened in step 502, the second end may extract at least two encryption weights corresponding to the screened at least two candidate leaf nodes from the encryption weights obtained in step 501, and sum the extracted at least two encryption weights to obtain the encryption score of the object to be processed in the tree model. Wherein the encryption score may be a score of the encrypted object to be processed. The score of the object to be processed may be a numerical value for characterizing the likelihood that the object to be processed belongs to a preset class. The higher the score, the greater the likelihood that the object to be processed belongs to the preset category may be characterized. The preset category may be a category that the tree model can predict.
In some alternative implementations of the present embodiment, the tree model includes at least two trees; and the second end may perform steps 502 and 503 described above by:
for a tree of at least two trees comprised by the tree model, the following steps are performed: screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree to serve as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree; and summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model.
The encrypted result is sent to the first end, step 504.
In this embodiment, based on the encryption score obtained in step 503, the second end may send the encryption score to the first end so that the first end classifies the object to be processed based on the encryption score.
It should be noted that the first end and the second end in the embodiment corresponding to fig. 5 may have similar or identical features to the first end and the second end in the embodiment corresponding to fig. 2, and the descriptions of the first end and the second end in the embodiment corresponding to fig. 2 also apply to the embodiment corresponding to fig. 5. In order to avoid repetition, the description is omitted here.
The method provided by the embodiment of the disclosure can be executed by the second end in the information processing system, the second end can complete score prediction for the object to be processed based on the encryption weight provided by the first end, and in the whole process, the second end only needs to send the obtained encryption result to the first end, and does not need to disclose the splitting direction of the object to be processed on the managed node to the first end, so that the privacy of the second end is protected, and the security of the data of the second end is improved.
With further reference to fig. 6, as an implementation of the method shown in the foregoing figures, the present disclosure provides an embodiment of an information processing apparatus, which corresponds to the method embodiment shown in fig. 4, and the apparatus may be configured at a first end in an information processing system, where the information processing system may include a first end and a second end, where the first end and the second end have the same tree model, and the first end and the second end respectively manage part of nodes in the tree model, and the first end has weights of leaf nodes of the tree model.
As shown in fig. 6, the information processing apparatus 600 of the present embodiment includes: a first acquiring unit 601 configured to acquire first feature information of an object to be processed; a first screening unit 602 configured to screen, as a target leaf node, a leaf node that a to-be-processed object does not reach from among leaf nodes included in the tree model based on the first feature information; an execution unit 603 configured to execute, for leaf nodes included in the tree model, the following processing steps: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; a first transmitting unit 604 configured to transmit the obtained encryption weight and a processing request for an object to be processed to the second end; and a receiving unit 605 configured to receive the encryption score from the second end, wherein the target encryption score is obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
In the present embodiment, the first acquiring unit 601 of the information processing apparatus 600 may acquire the first characteristic information of the object to be processed from a remote place or a local place by a wired connection manner or a wireless connection manner. The object to be processed may be an object to be processed, and specifically may be various objects that can be processed by the tree model. The first characteristic information of the object to be processed may be used to characterize the characteristics of the object to be processed. Specifically, the characteristic represented by the first characteristic information acquired by the first end may be the same as the characteristic corresponding to the node managed by the first end.
In this embodiment, based on the first feature information obtained by the first obtaining unit 601, the first screening unit 602 may screen, as the target leaf node, a leaf node that the object to be processed will not reach from among leaf nodes included in the tree model.
In the present embodiment, the execution unit 603 may execute the following processing steps (step 6031-step 6032) for the leaf nodes included in the tree model:
in step 6031, in response to the leaf node being the target leaf node, encrypting 0 to obtain the encryption weight corresponding to the target leaf node.
In step 6032, in response to the leaf node not being the target leaf node, the weight of the leaf node is encrypted, and the encryption weight corresponding to the leaf node is obtained.
In the present embodiment, based on the encryption weight obtained by the execution unit 603, the first transmission unit 604 may transmit the encryption weight to the second terminal, and transmit a processing request for the above-described object to be processed to the second terminal. Wherein the processing request may be used to request processing of the object to be processed.
In this embodiment, after transmitting the encryption weight and the processing request to the second end, the receiving unit 605 of the apparatus 600 may receive the encryption score generated based on the encryption weight from the second end. Wherein the encryption score may be a score of the encrypted object to be processed. The score of the object to be processed may be a numerical value for characterizing the likelihood that the object to be processed belongs to a preset class. The higher the score, the greater the likelihood that the object to be processed belongs to the preset category may be characterized. The preset category may be a category that the tree model can predict.
Specifically, the encryption score may be obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model as at least two candidate leaf nodes based on the second characteristic information; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
In some optional implementations of this embodiment, the apparatus 600 may further include: a decryption unit (not shown in the figure) configured to decrypt the received encryption score to obtain a target score; a classification unit (not shown in the figure) configured to classify the object to be processed based on the target score.
In some optional implementations of the present embodiment, the execution unit 603 is further configured to: in response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node; and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node. It will be appreciated that the elements described in the apparatus 600 correspond to the various steps in the method described in fig. 4. Thus, the operations, features and resulting benefits described above with respect to the method are equally applicable to the apparatus 600 and the units contained therein, and are not described in detail herein.
The device 600 provided in the foregoing embodiment of the present disclosure may be configured on the first end, and further configured with the first end of the device 600 to encrypt the virtual weight (i.e. 0) of the target leaf node and the real weight of the leaf node other than the target leaf node, and then send the encrypted virtual weight and the real weight of the leaf node other than the target leaf node to the second end, where the second end cannot distinguish the target leaf node, and further cannot learn the splitting direction of the first end, and the encryption of 0 may further enable the contribution degree of the target leaf node to the score of the object to be processed to be 0, and further may ensure the accuracy of the score of the object to be processed obtained by summing.
With further reference to fig. 7, as an implementation of the method shown in the foregoing figures, the present disclosure provides another embodiment of an information processing apparatus, where the apparatus embodiment corresponds to the method embodiment shown in fig. 5, and the apparatus may be configured on a second end in an information processing system, where the information processing system may include a first end and a second end, where the first end and the second end have the same tree model, and the first end and the second end respectively manage part of nodes in the tree model, and the first end has weights of leaf nodes of the tree model.
As shown in fig. 7, the information processing apparatus 700 of the present embodiment includes: a second obtaining unit 701 configured to obtain second feature information of an object to be processed in response to receiving an encryption weight sent by a first end and a processing request for the object to be processed, wherein the encryption weight is obtained by the first end by: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; a second screening unit 702 configured to screen, based on the second feature information, at least two leaf nodes that may be reached by the object to be processed from among leaf nodes included in the tree model as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes include a target leaf node; a summing unit 703 configured to sum the encryption weights corresponding to the at least two candidate leaf nodes to obtain an encryption score of the object to be processed in the tree model; the second transmitting unit 704 is configured to transmit the encrypted result to the first end.
In this embodiment, the second obtaining unit 701 of the information processing apparatus 700 may obtain the second feature information of the object to be processed in response to receiving the encryption weight sent by the first end and the processing request for the object to be processed through the wired connection manner or the wireless connection manner. The obtained encryption weights may include encryption weights corresponding to respective leaf nodes in the tree model. The second characteristic information of the object to be processed may be used to characterize the characteristics of the object to be processed. Specifically, the characteristic represented by the second characteristic information acquired by the second end may be the same as the characteristic corresponding to the node managed by the second end. The processing request may be used to request processing of the object to be processed.
In this embodiment, the encryption weight may be obtained by the first end by: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node.
In this embodiment, based on the second feature information obtained by the second obtaining unit 701, the second screening unit 702 may screen at least two leaf nodes that may be reached by the object to be processed from leaf nodes included in the tree model as at least two candidate leaf nodes. Wherein the at least two candidate leaf nodes may include a target leaf node screened out by the first end.
In this embodiment, based on the at least two candidate leaf nodes screened by the second screening unit 702, the summing unit 703 may extract at least two encryption weights corresponding to the screened at least two candidate leaf nodes from the encryption weights obtained by the second obtaining unit 701, and sum the extracted at least two encryption weights to obtain an encryption score of the object to be processed in the tree model. Wherein the encryption score may be a score of the encrypted object to be processed. The score of the object to be processed may be a numerical value for characterizing the likelihood that the object to be processed belongs to a preset class. The higher the score, the greater the likelihood that the object to be processed belongs to the preset category may be characterized. The preset category may be a category that the tree model can predict.
In the present embodiment, based on the encryption score obtained by the summing unit 703, the second transmitting unit 704 may transmit the encryption score to the first end so that the first end classifies the object to be processed based on the encryption score.
It will be appreciated that the elements described in the apparatus 700 correspond to the various steps in the method described in fig. 5. Thus, the operations, features and resulting benefits described above for the method are equally applicable to the apparatus 700 and the units contained therein, and are not described in detail herein.
In some alternative implementations of the present embodiment, the tree model includes at least two trees; and the second screening unit 702 and the summing unit 703 are further configured to: for a tree of at least two trees comprised by the tree model, the following steps are performed: screening at least two leaf nodes which are possibly reached by an object to be processed from leaf nodes included in the tree as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree; and summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model. The device 700 provided in the foregoing embodiment of the present disclosure may be configured on the second end, and further, the second end configured with the device 700 may complete score prediction for the object to be processed based on the encryption weight provided by the first end, and in the whole process, the second end only needs to send the obtained encryption result to the first end, and does not need to disclose the splitting direction of the object to be processed on the managed node to the first end, which is helpful for protecting the privacy of the second end and improving the security of the data of the second end.
Referring now to fig. 8, a schematic diagram of an electronic device (e.g., server in fig. 1) 800 suitable for use in implementing embodiments of the present disclosure is shown. The terminal devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and the like, and stationary terminals such as digital TVs, desktop computers, and the like. The electronic device shown in fig. 8 is merely an example and should not be construed to limit the functionality and scope of use of the disclosed embodiments.
As shown in fig. 8, the electronic device 800 may include a processing means (e.g., a central processor, a graphics processor, etc.) 801, which may perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 802 or a program loaded from a storage means 808 into a Random Access Memory (RAM) 803. In the RAM803, various programs and data required for the operation of the electronic device 800 are also stored. The processing device 801, the ROM 802, and the RAM803 are connected to each other by a bus 804. An input/output (I/O) interface 805 is also connected to the bus 804.
In general, the following devices may be connected to the I/O interface 805: input devices 806 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, and the like; an output device 807 including, for example, a Liquid Crystal Display (LCD), speakers, vibrators, etc.; storage 808 including, for example, magnetic tape, hard disk, etc.; communication means 809. The communication means 809 may allow the electronic device 800 to communicate wirelessly or by wire with other devices to exchange data. While fig. 8 shows an electronic device 800 having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead.
In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network via communication device 809, or installed from storage device 808, or installed from ROM 802. The above-described functions defined in the methods of the embodiments of the present disclosure are performed when the computer program is executed by the processing device 801.
It should be noted that the computer readable medium described in the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present disclosure, however, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.
The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; sending the obtained encryption weight and a processing request for the object to be processed to a second end; and receiving an encryption score from the second end, wherein the target encryption score is obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
Alternatively, the electronic device may be further caused to: in response to receiving the encryption weight sent by the first end and the processing request aiming at the object to be processed, obtaining second characteristic information of the object to be processed, wherein the encryption weight is obtained by the first end through the following steps: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which are not reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; screening at least two leaf nodes which are possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model; the encrypted result is sent to the first end.
Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units involved in the embodiments described in the present disclosure may be implemented by means of software, or may be implemented by means of hardware. The described units may also be provided in a processor, for example, described as: a processor includes a first acquisition unit, a first screening unit, an execution unit, a first transmission unit, and a reception unit. The names of these units do not constitute a limitation on the unit itself in some cases, and for example, the receiving unit may also be described as "a unit that receives an encryption score".
The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by persons skilled in the art that the scope of the disclosure referred to in this disclosure is not limited to the specific combinations of features described above, but also covers other embodiments which may be formed by any combination of features described above or equivalents thereof without departing from the spirit of the disclosure. Such as those described above, are mutually substituted with the technical features having similar functions disclosed in the present disclosure (but not limited thereto).
Claims (15)
1. An information handling system comprising a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing a portion of nodes in the tree model, respectively, the first end having weights for leaf nodes of the tree model, wherein,
The first end is configured to: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which cannot be reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node; sending the obtained encryption weight and a processing request aiming at the object to be processed to the second end;
the second end is configured to: acquiring second characteristic information of the object to be processed in response to receiving the encryption weight sent by the first end and a processing request for the object to be processed; screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain encryption scores of the object to be processed in the tree model; and sending the encrypted result to the first end.
2. The system of claim 1, wherein the tree model comprises at least two trees; and
the second end is further configured to:
for a tree of at least two trees comprised by the tree model, performing the steps of: screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree to serve as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree;
and summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model.
3. The system of claim 1 or 2, wherein the first end is further configured to:
decrypting the received encrypted score to obtain a target score;
and classifying the object to be processed based on the target score.
4. The system of claim 1 or 2, wherein the first end is further configured to:
in response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node;
And in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node.
5. An information processing method performed by a first end in an information processing system, the information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing part of nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the method comprising:
acquiring first characteristic information of an object to be processed;
based on the first characteristic information, screening leaf nodes which cannot be reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes;
for leaf nodes included in the tree model, the following processing steps are performed:
in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node;
in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node;
Sending the obtained encryption weight and a processing request aiming at the object to be processed to the second end; and
receiving an encryption score from the second end, wherein the target encryption score is obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
6. The method of claim 5, wherein the method further comprises:
decrypting the received encrypted score to obtain a target score;
and classifying the object to be processed based on the target score.
7. The method according to claim 5 or 6, wherein, in response to the leaf node being a target leaf node, encrypting 0 to obtain an encryption weight corresponding to the target leaf node includes:
In response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node; and
and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node, wherein obtaining the encryption weight corresponding to the leaf node comprises:
and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node.
8. An information processing method performed by a second end in an information processing system, the information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing part of nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the method comprising:
in response to receiving the encryption weight sent by the first end and the processing request aiming at the object to be processed, obtaining second characteristic information of the object to be processed, wherein the encryption weight is obtained by the first end through the following steps: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which cannot be reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node;
Screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes;
summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain encryption scores of the object to be processed in the tree model;
and sending the encrypted result to the first end.
9. The method of claim 8, wherein the tree model comprises at least two trees; and
screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information to serve as at least two candidate leaf nodes; summing the encryption weights corresponding to the at least two candidate leaf nodes, wherein obtaining the encryption score of the object to be processed in the tree model comprises:
for a tree of at least two trees comprised by the tree model, performing the steps of: screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree to serve as at least two candidate leaf nodes; summing the encryption weights corresponding to at least two candidate leaf nodes of the tree to obtain an initial encryption score corresponding to the tree;
And summing at least two initial encryption scores corresponding to at least two trees included in the tree model to obtain the encryption score of the object to be processed in the tree model.
10. An information processing apparatus configured at a first end in an information processing system, the information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing part of nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the apparatus comprising:
a first acquisition unit configured to acquire first characteristic information of an object to be processed;
a first screening unit configured to screen, based on the first feature information, a leaf node, which the object to be processed does not reach, from among leaf nodes included in the tree model as a target leaf node;
an execution unit configured to execute, for leaf nodes included in the tree model, the following processing steps:
in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node;
in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node;
A first transmitting unit configured to transmit the obtained encryption weight and a processing request for the object to be processed to the second end; and
a receiving unit configured to receive an encryption score from the second end, wherein the target encryption score is obtained by the second end by: acquiring second characteristic information of an object to be processed; screening at least two leaf nodes possibly reached by the object to be processed from leaf nodes included in the tree model based on the second characteristic information as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes comprise target leaf nodes; and summing the encryption weights corresponding to the at least two candidate leaf nodes to obtain the encryption score of the object to be processed in the tree model.
11. The apparatus of claim 10, wherein the apparatus method further comprises:
a decryption unit configured to decrypt the received encryption score to obtain a target score;
and a classification unit configured to classify the object to be processed based on the target score.
12. The apparatus of claim 10 or 11, wherein the execution unit is further configured to:
In response to the leaf node being a target leaf node, encrypting 0 in a homomorphic encryption mode to obtain an encryption weight corresponding to the target leaf node;
and in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node in a homomorphic encryption mode to obtain the encryption weight corresponding to the leaf node.
13. An information processing apparatus configured at a second end in an information processing system, the information processing system including a first end and a second end, the first end and the second end having a same tree model, the first end and the second end managing partial nodes in the tree model, respectively, the first end having weights of leaf nodes of the tree model, the apparatus comprising:
a second obtaining unit configured to obtain second feature information of an object to be processed in response to receiving an encryption weight sent by a first end and a processing request for the object to be processed, wherein the encryption weight is obtained by the first end by: acquiring first characteristic information of an object to be processed; based on the first characteristic information, screening leaf nodes which cannot be reached by the object to be processed from leaf nodes included in the tree model to serve as target leaf nodes; for leaf nodes included in the tree model, the following processing steps are performed: in response to the leaf node being a target leaf node, encrypting 0 to obtain encryption weight corresponding to the target leaf node; in response to the leaf node not being the target leaf node, encrypting the weight of the leaf node to obtain the encryption weight corresponding to the leaf node;
A second screening unit configured to screen, based on the second feature information, at least two leaf nodes that the object to be processed may reach from among leaf nodes included in the tree model as at least two candidate leaf nodes, wherein the at least two candidate leaf nodes include a target leaf node;
a summing unit configured to sum the encryption weights corresponding to the at least two candidate leaf nodes to obtain an encryption score of the object to be processed in the tree model;
a second transmitting unit configured to transmit the encrypted result to the first end.
14. An electronic device, comprising:
one or more processors;
a storage device having one or more programs stored thereon,
when executed by the one or more processors, causes the one or more processors to implement the method of any of claims 5-9.
15. A computer readable medium having stored thereon a computer program, wherein the program when executed by a processor implements the method of any of claims 5-9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011018697.7A CN113807530B (en) | 2020-09-24 | 2020-09-24 | Information processing system, method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011018697.7A CN113807530B (en) | 2020-09-24 | 2020-09-24 | Information processing system, method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113807530A CN113807530A (en) | 2021-12-17 |
CN113807530B true CN113807530B (en) | 2024-02-06 |
Family
ID=78943495
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011018697.7A Active CN113807530B (en) | 2020-09-24 | 2020-09-24 | Information processing system, method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113807530B (en) |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007189597A (en) * | 2006-01-16 | 2007-07-26 | Univ Of Electro-Communications | Encryption device, encryption method, decoding device, and decoding method |
CN102843372A (en) * | 2012-08-28 | 2012-12-26 | 西安交通大学 | Order-preserving encryption method based on random interval partition |
WO2018122238A1 (en) * | 2016-12-30 | 2018-07-05 | Robert Bosch Gmbh | Method and system for fuzzy keyword search over encrypted data |
CN110138802A (en) * | 2019-06-02 | 2019-08-16 | 四川虹微技术有限公司 | User's characteristic information acquisition methods, device, block chain node and storage medium |
CN110457912A (en) * | 2019-07-01 | 2019-11-15 | 阿里巴巴集团控股有限公司 | Data processing method, device and electronic equipment |
CN110874479A (en) * | 2018-08-29 | 2020-03-10 | 阿里巴巴集团控股有限公司 | Method and system for safely processing decision tree model, data terminal and processing terminal |
CN110944011A (en) * | 2019-12-16 | 2020-03-31 | 支付宝(杭州)信息技术有限公司 | Joint prediction method and system based on tree model |
CN111046408A (en) * | 2019-12-13 | 2020-04-21 | 支付宝(杭州)信息技术有限公司 | Judgment result processing method, query method, device, electronic equipment and system |
CN111695697A (en) * | 2020-06-12 | 2020-09-22 | 深圳前海微众银行股份有限公司 | Multi-party combined decision tree construction method and device and readable storage medium |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11196541B2 (en) * | 2017-01-20 | 2021-12-07 | Enveil, Inc. | Secure machine learning analytics using homomorphic encryption |
US20200175426A1 (en) * | 2019-07-01 | 2020-06-04 | Alibaba Group Holding Limited | Data-based prediction results using decision forests |
-
2020
- 2020-09-24 CN CN202011018697.7A patent/CN113807530B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007189597A (en) * | 2006-01-16 | 2007-07-26 | Univ Of Electro-Communications | Encryption device, encryption method, decoding device, and decoding method |
CN102843372A (en) * | 2012-08-28 | 2012-12-26 | 西安交通大学 | Order-preserving encryption method based on random interval partition |
WO2018122238A1 (en) * | 2016-12-30 | 2018-07-05 | Robert Bosch Gmbh | Method and system for fuzzy keyword search over encrypted data |
CN110874479A (en) * | 2018-08-29 | 2020-03-10 | 阿里巴巴集团控股有限公司 | Method and system for safely processing decision tree model, data terminal and processing terminal |
CN110138802A (en) * | 2019-06-02 | 2019-08-16 | 四川虹微技术有限公司 | User's characteristic information acquisition methods, device, block chain node and storage medium |
CN110457912A (en) * | 2019-07-01 | 2019-11-15 | 阿里巴巴集团控股有限公司 | Data processing method, device and electronic equipment |
CN111046408A (en) * | 2019-12-13 | 2020-04-21 | 支付宝(杭州)信息技术有限公司 | Judgment result processing method, query method, device, electronic equipment and system |
CN110944011A (en) * | 2019-12-16 | 2020-03-31 | 支付宝(杭州)信息技术有限公司 | Joint prediction method and system based on tree model |
CN111695697A (en) * | 2020-06-12 | 2020-09-22 | 深圳前海微众银行股份有限公司 | Multi-party combined decision tree construction method and device and readable storage medium |
Non-Patent Citations (1)
Title |
---|
云计算环境下基于密文策略的权重属性加密方案;刘西蒙;马建峰;熊金波;李琦;张涛;朱辉;;四川大学学报(工程科学版)(06);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN113807530A (en) | 2021-12-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110245510B (en) | Method and apparatus for predicting information | |
CN112784823B (en) | Face image recognition method, face image recognition device, computing equipment and medium | |
CN110084317B (en) | Method and device for recognizing images | |
CN112434620B (en) | Scene text recognition method, device, equipment and computer readable medium | |
CN117633228A (en) | Model training method and device | |
CN110765490A (en) | Method and apparatus for processing information | |
CN110008926B (en) | Method and device for identifying age | |
CN113722738B (en) | Data protection method, device, medium and electronic equipment | |
CN114006769A (en) | Model training method and device based on horizontal federal learning | |
US10049222B1 (en) | Establishing application trust levels using taint propagation | |
CN110705635B (en) | Method and apparatus for generating an isolated forest | |
CN113807530B (en) | Information processing system, method and device | |
US20230418794A1 (en) | Data processing method, and non-transitory medium and electronic device | |
CN116361121A (en) | Abnormal interface alarm method, device, electronic equipment and computer readable medium | |
CN111680754B (en) | Image classification method, device, electronic equipment and computer readable storage medium | |
CN114595474A (en) | Federal learning modeling optimization method, electronic device, medium, and program product | |
CN110392064B (en) | Risk identification method and device, computing equipment and computer readable storage medium | |
CN109614137B (en) | Software version control method, device, equipment and medium | |
CN110991312A (en) | Method, apparatus, electronic device, and medium for generating detection information | |
CN111752625A (en) | Method and device for interface mock | |
CN118214617B (en) | Network security protection method, device, electronic equipment and medium | |
CN114513548B (en) | Directional call information processing method and device | |
CN116226888B (en) | Power data interactive encryption method, system and equipment based on privacy protection | |
CN111786937B (en) | Method, apparatus, electronic device and readable medium for identifying malicious request | |
CN116843800B (en) | Animation information transmission method, device, electronic equipment and computer readable medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |