Big data risk control system based on federal learning
By adopting a federated learning-based architecture, combined with lightweight deep neural networks and privacy protection technologies, the problems of data silos and privacy leaks in traditional big data risk control systems are solved. This enables the aggregation of risk features across institutions and the sharing of privacy data, thereby improving the security and efficiency of the risk control system.
Patent Information
- Application Number
- CN202510680540.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-26
- Publication Date
- 2025-10-21
AI Technical Summary
Traditional big data risk control systems rely on centralized data warehouses, which suffer from data silos, privacy risks, and insufficient model generalization capabilities. Furthermore, they are prone to leaking personal information in cross-institutional collaborations and lag behind in model iteration when facing new and complex fraud methods.
It adopts a federated learning-based architecture, including a local federated learning module, an anonymized data processing engine, and a cloud-based collaborative training platform. It combines lightweight deep neural networks, differential privacy and homomorphic encryption, dynamic data obfuscation algorithms, federated feature embedding, edge computing, and other technologies to achieve distributed training and privacy protection, and supports cross-institutional risk feature aggregation and privacy data sharing.
It enables the aggregation of risk features across institutions and the sharing of privacy data with zero leakage, improves the generalization ability of the model and the real-time response to new fraud patterns, ensures privacy and security and the auditability of the training process, and enhances the security and efficiency of the risk control system.
Smart Images
Figure CN120822239A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of self-service terminal technology, and in particular to a big data risk control system based on federated learning. Background Art
[0002] Traditional big data risk control systems rely on centralized data warehouses, and have defects such as data silos, privacy leakage risks and insufficient model generalization capabilities; existing technologies require sharing original data in plain text in cross-institutional collaboration, which is prone to leaking personal information, and model iteration lags behind when faced with new and complex fraud methods. Summary of the Invention
[0003] The purpose of the present invention is to provide a big data risk control system based on federated learning to address the above-mentioned problems in the prior art, thereby solving all or one of the above-mentioned problems in the prior art.
[0004] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows: The present invention provides a big data risk control system based on federated learning, comprising: A local federated learning module, an anonymized data processing engine, and a cloud-based collaborative training platform are used to conduct distributed training of anti-fraud models through terminal devices, thereby achieving cross-institutional risk feature aggregation and zero-leakage sharing of privacy data.
[0005] As an improved solution, a lightweight deep neural network is deployed in the local federated learning module, which is used to use a hybrid technology of differential privacy and homomorphic encryption to complete user behavior feature extraction and gradient desensitization processing on the terminal device.
[0006] As an improved solution, the anonymized data processing engine is integrated with a dynamic data obfuscation algorithm, which is used to construct synthetic data samples through a generative adversarial network, retaining the risk pattern of the original data while eliminating individual identity identifiers.
[0007] As an improved solution, the cloud-based collaborative training platform is used to build a cross-institutional federal alliance chain, supporting the secure aggregation of multi-node model parameters and alignment of heterogeneous data features, and realizing incremental updates of the joint risk rule base.
[0008] As an improved solution, the big data risk control system based on federated learning is also equipped with a dynamic model optimization engine, which is used to capture new fraud patterns in real time through an online meta-learning algorithm and automatically trigger local fine-tuning of the federated model and global parameter synchronization.
[0009] As an improved solution, the big data risk control system based on federated learning is also equipped with a real-time abnormal transaction interception engine, which is used to deploy a graph neural network based on federated feature embedding and achieve millisecond-level risk decision-making through high-risk transaction subgraph clustering analysis.
[0010] As an improved solution, the big data risk control system based on federated learning is also equipped with a privacy protection enhancement mechanism, which is used to adopt a dual isolation strategy of federated differential privacy and trusted execution environment to ensure the auditability and irreversibility of intermediate parameters during model training.
[0011] As an improved solution, the big data risk control system based on federated learning is also equipped with cross-institutional data federation, which is used to implement indexing and retrieval of heterogeneous data sources through distributed hash tables, and supports the linkage of multi-scenario risk rules.
[0012] As an improved solution, the big data risk control system based on federated learning is also equipped with an edge computing optimization engine, which is used to dynamically adjust the federated learning batch size and encryption strength according to the computing power of the terminal device, thereby improving the training convergence speed while ensuring privacy and security.
[0013] As an improved solution, the big data risk control system based on federated learning is also provided with a feedback learning mechanism, which is used to continuously enhance the federated model's resistance to adversarial sample attacks through adversarial perturbation generation and model robustness verification.
[0014] The beneficial effect of the technical solution of the present invention is: the present invention constructs a distributed risk control architecture that integrates federated learning and privacy computing, innovatively integrates dynamic data obfuscation, federated differential privacy and edge computing optimization technologies, and breaks through the data barriers and security compliance bottlenecks of traditional risk control systems. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0016] Figure 1 2 is a schematic diagram of the architecture of a big data risk control system based on federated learning according to an embodiment of the present invention. DETAILED DESCRIPTION
[0017] The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more precise definition of the protection scope of the present invention.
[0018] In the description of the present invention, it should be noted that the embodiments described in the present invention are only part of the embodiments of the present invention, rather than all of the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention.
[0019] The terms "first," "second," and the like in the specification and claims herein and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments of the present invention described herein can be implemented in orders other than those illustrated or described herein. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or device.
[0020] This embodiment provides a big data risk control system based on federated learning. Figure 1 Shown, including: A local federated learning module, an anonymized data processing engine, and a cloud-based collaborative training platform are used to conduct distributed training of anti-fraud models through terminal devices, thereby achieving cross-institutional risk feature aggregation and zero-leakage sharing of privacy data.
[0021] As an improved solution, a lightweight deep neural network is deployed in the local federated learning module, which is used to use a hybrid technology of differential privacy and homomorphic encryption to complete user behavior feature extraction and gradient desensitization processing on the terminal device.
[0022] As an improved solution, the anonymized data processing engine is integrated with a dynamic data obfuscation algorithm, which is used to construct synthetic data samples through a generative adversarial network, retaining the risk pattern of the original data while eliminating individual identity identifiers.
[0023] As an improved solution, the cloud-based collaborative training platform is used to build a cross-institutional federal alliance chain, supporting the secure aggregation of multi-node model parameters and alignment of heterogeneous data features, and realizing incremental updates of the joint risk rule base.
[0024] As an improved solution, the big data risk control system based on federated learning is also equipped with a dynamic model optimization engine, which is used to capture new fraud patterns in real time through an online meta-learning algorithm and automatically trigger local fine-tuning of the federated model and global parameter synchronization.
[0025] As an improved solution, the big data risk control system based on federated learning is also equipped with a real-time abnormal transaction interception engine, which is used to deploy a graph neural network based on federated feature embedding and achieve millisecond-level risk decision-making through high-risk transaction subgraph clustering analysis.
[0026] As an improved solution, the big data risk control system based on federated learning is also equipped with a privacy protection enhancement mechanism, which is used to adopt a dual isolation strategy of federated differential privacy and trusted execution environment to ensure the auditability and irreversibility of intermediate parameters during model training.
[0027] As an improved solution, the big data risk control system based on federated learning is also equipped with cross-institutional data federation, which is used to implement indexing and retrieval of heterogeneous data sources through distributed hash tables, and supports the linkage of multi-scenario risk rules.
[0028] As an improved solution, the big data risk control system based on federated learning is also equipped with an edge computing optimization engine, which is used to dynamically adjust the federated learning batch size and encryption strength according to the computing power of the terminal device, thereby improving the training convergence speed while ensuring privacy and security.
[0029] As an improved solution, the big data risk control system based on federated learning is also provided with a feedback learning mechanism, which is used to continuously enhance the federated model's resistance to adversarial sample attacks through adversarial perturbation generation and model robustness verification.
[0030] It should be noted that the above examples are only for explaining the present invention and are not intended to limit the scope of protection of the present invention.
[0031] It should also be understood that in the embodiments herein, the term "and / or" merely describes an association between associated objects, indicating that three possible relationships exist. For example, "A and / or B" could represent: A alone, A and B together, or B alone. Furthermore, the character " / " in this document generally indicates an "or" relationship between the associated objects.
[0032] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this document.
[0033] In the several embodiments provided herein, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices, or units, or can be an electrical, mechanical, or other form of connection.
[0034] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments herein.
[0035] In addition, the functional units in the various embodiments herein may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.
[0036] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this article is essentially or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of this article. The aforementioned storage medium includes: various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0037] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention's description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A big data risk control system based on federated learning, characterized by: include: A local federated learning module, an anonymized data processing engine, and a cloud-based collaborative training platform are used to conduct distributed training of anti-fraud models through terminal devices, thereby achieving cross-institutional risk feature aggregation and zero-leakage sharing of privacy data.
2. The big data risk control system based on federated learning according to claim 1, characterized in that: A lightweight deep neural network is deployed in the local federated learning module, which is used to use a hybrid technology of differential privacy and homomorphic encryption to complete user behavior feature extraction and gradient desensitization processing on the terminal device.
3. The big data risk control system based on federated learning according to claim 1, characterized in that: The anonymized data processing engine is integrated with a dynamic data obfuscation algorithm, which is used to construct synthetic data samples through a generative adversarial network, preserving the risk patterns of the original data while eliminating individual identity identifiers.
4. The big data risk control system based on federated learning according to claim 1, characterized in that: The cloud-based collaborative training platform is used to build a cross-institutional federal alliance chain, support the secure aggregation of multi-node model parameters and alignment of heterogeneous data features, and realize incremental updates of the joint risk rule base.
5. The big data risk control system based on federated learning according to claim 1, characterized in that: The big data risk control system based on federated learning is also equipped with a dynamic model optimization engine, which is used to capture new fraud patterns in real time through an online meta-learning algorithm and automatically trigger local fine-tuning of the federated model and global parameter synchronization.
6. The big data risk control system based on federated learning according to claim 1, characterized in that: The big data risk control system based on federated learning is also equipped with a real-time abnormal transaction interception engine, which is used to deploy a graph neural network based on federated feature embedding and realize millisecond-level risk decision-making through high-risk transaction subgraph clustering analysis.
7. The big data risk control system based on federated learning according to claim 1, characterized in that: The big data risk control system based on federated learning is also equipped with a privacy protection enhancement mechanism, which is used to adopt a dual isolation strategy of federated differential privacy and trusted execution environment to ensure the auditability and irreversibility of intermediate parameters during model training.
8. The big data risk control system based on federated learning according to claim 1, characterized in that: The big data risk control system based on federated learning is also equipped with cross-institutional data federation, which is used to implement indexing and retrieval of heterogeneous data sources through distributed hash tables, and supports the linkage of multi-scenario risk rules.
9. The big data risk control system based on federated learning according to claim 1, characterized in that: The big data risk control system based on federated learning is also equipped with an edge computing optimization engine, which is used to dynamically adjust the federated learning batch size and encryption strength according to the computing power of the terminal device, thereby improving the training convergence speed while ensuring privacy and security.
10. The big data risk control system based on federated learning according to claim 1, characterized in that: The big data risk control system based on federated learning is also provided with a feedback learning mechanism, which is used to continuously enhance the federated model's resistance to adversarial sample attacks through adversarial perturbation generation and model robustness verification.