Sensitive Data Leakage Detection Using GAN and Transformer Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Microservice architectures in enterprise computing systems face challenges with sensitive data leakage due to high false positive rates and latency, as existing methods like regular expression matching lack nuance and precision, leading to inefficient resource usage.
Innovation Solution
Implementing generative adversarial networks (GANs) and transformer models to detect and prevent sensitive data leakage by applying a blocking policy, reducing false positives and enhancing system performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If regular expression matching is used to detect sensitive data, then the implementation is simple, but the false positive rate is high and precision is low
Solution Approach 1:
The patent replaces traditional mechanical regular expression matching with a neural network-based machine learning system. The neural network model learns complex patterns of sensitive data through training, enabling accurate detection without relying on rigid pattern matching rules. This substitution resolves the contradiction by achieving high detection precision while maintaining implementation feasibility through automated model training.
Solution Approach 2:
The patent transforms the detection approach from static pattern matching to dynamic machine learning classification. By changing the parameter representation from fixed regular expression patterns to learned neural network weights and biases, the system achieves adaptive detection that reduces false positives while maintaining simplicity through automated parameter optimization during training.
2Device complexity
If traditional methods are used for sensitive data detection, then the system complexity is low, but the resource usage is inefficient due to high false positive rates
Solution Approach 1:
The patent implements preliminary training of neural network models using labeled sensitive data before deployment. This preliminary action creates an optimized detection system that requires minimal runtime resources. The model learns detection patterns in advance, enabling efficient real-time detection without requiring complex runtime analysis, thus resolving the contradiction between system complexity and resource efficiency.
Solution Approach 2:
The patent uses copied and transformed versions of sensitive data during training to teach the neural network detection patterns. By creating training samples from actual sensitive data through various transformations, the system learns to recognize patterns efficiently without requiring complex runtime processing, improving resource utilization while maintaining manageable system complexity.
3Adaptability or versatility
If microservice architectures are implemented, then the system flexibility and modularity are improved, but the risk of sensitive data leakage increases due to numerous data transmission endpoints
Solution Approach 1:
The patent implements a universal neural network-based detection model that can be deployed across multiple microservice endpoints. This single model handles diverse detection requirements for different types of sensitive data (PII, financial information, healthcare data) across all services, reducing the overall risk by providing consistent protection throughout the distributed system without requiring service-specific customization.
Solution Approach 2:
The patent introduces a centralized sensitive data detection service as an intermediary between microservices and data transmission endpoints. This mediator intercepts and analyzes data flows across all service boundaries, providing coordinated protection that reduces leakage risk while preserving the flexibility and modularity of the microservice architecture through a unified detection layer.
Data Source
AI summary
Disclosed herein are system, method, and computer program product embodiments for increasing data security by using generative adversarial networks (GAN) and transformer models to detect sensitive data leakage. A transformer model may receive a message via a network. The transformer model may then apply a GAN model to determine whether the message contains potentially sensitive data requiring further inspection. If the message contains sensitive data, a blocking policy may be applied to discard the message or remove sensitive data from the message prior to transmission of the message, thereby preventing sensitive data leakage.