Encrypted Decision Tree Generation via Base-Point Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data analysis methods, such as decision tree analysis, face challenges in maintaining secrecy of sensitive information during the learning phase, especially when using external data analysis services.
Innovation Solution
A data analysis server employs high-performance encryption to generate and process base-point-added information, associating encrypted explanatory and response variables with base points, allowing decision tree generation without decrypting the variables, thus enhancing data secrecy during the learning phase.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If data analysis is performed using external services, then analysis capability is improved, but data secrecy deteriorates
Solution Approach 1:
The patent segments the decision tree generation process into multiple stages: generating branching rule candidates using encrypted data, evaluating these candidates using base-point-added information, and selecting the best candidate. This segmentation allows external services to participate in analysis while keeping sensitive data encrypted throughout the process.
Solution Approach 2:
The patent introduces base-point-added information as an intermediary element that enables external analysis services to evaluate branching rules without accessing the actual encrypted data. This intermediary structure allows the service provider to perform analysis tasks while the data owner maintains control over data secrecy through encryption keys.
2Loss of information
If encryption is applied to protect data secrecy, then data secrecy is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-calculating and storing base-point-added information during the learning phase. This pre-computed information is used later to evaluate branching rule candidates without requiring decryption, thereby reducing processing complexity during actual analysis while maintaining data secrecy through encryption.
3Loss of information
If encrypted data is used for decision tree learning, then data secrecy is improved, but learning efficiency deteriorates
Solution Approach 1:
The patent applies partial decryption or evaluation by using base-point-added information that contains aggregated statistical properties of the encrypted data. Instead of fully decrypting data for analysis, the system performs partial operations on encrypted representations, achieving sufficient learning efficiency while maintaining data secrecy.
Data Source
AI summary
A data analysis server holds base-point-added information, wherein the base-point-added information includes a value of an explanatory variable encrypted by first high-performance encryption, a value of a response variable encrypted by a predetermined encryption scheme, and a base point based on frequencies of the value of the explanatory variable in information for learning are associated with one another, wherein the data analysis server: executes decision tree generation processing for generating a decision tree having a leaf node associated with the value of the response variable encrypted by the predetermined encryption scheme, which is included in the base-point-added information; and execute branching rule determine processing in decision tree generation processing without decrypting the value of the explanatory variable and the base point corresponding.