Private Categorization Using Shared Keys for Encrypted Transaction Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge is to preserve user privacy while combining data from multiple users to train machine learning models, as using data from multiple users increases the risk of exposing private information.
Innovation Solution
The method employs private categorization using shared keys, where users encrypt their transaction data with a shared key, and machine learning models are trained using homomorphically encrypted information, preventing access to unencrypted data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data from multiple users is combined to train machine learning models, then model training quality is improved, but user privacy is compromised
Solution Approach 1:
The patent introduces homomorphic encryption as an intermediary mechanism that allows the classifier model to process encrypted transaction vectors without decrypting them. The encryption key acts as a mediator between the need for data processing (to improve model quality) and the need to preserve privacy (by keeping data encrypted throughout training). This resolves the contradiction by enabling model training on aggregated multi-user data while maintaining cryptographic protection of individual user information.
Solution Approach 2:
The patent applies different cryptographic operations to different parts of the data processing pipeline. Transaction vectors are encrypted with specific keys before being pooled with other users' data, while the classifier model uses corresponding decryption keys to process only the portions of data it needs. This localized application of encryption/decryption to specific data elements and processing stages allows model training quality to improve through data aggregation while privacy is preserved through selective cryptographic protection.
2Object-affected harmful factors
If encryption is applied to transaction data, then user privacy is preserved, but processing complexity increases
Solution Approach 1:
The patent applies encryption to transaction vectors before they are pooled and processed by the classifier model. This preliminary encryption action ensures that privacy protection is established prior to any processing operations, eliminating the need for complex secure computation protocols during the actual training phase. The preprocessing step of encryption simplifies subsequent processing by allowing the model to work with encrypted data directly, reducing overall system complexity.
Data Source
AI summary
A method implements private categorization using shared keys. The method includes selecting an encryption key, encrypting a transaction vector, generated from a transaction record, with the encryption key to generate an encrypted transaction vector, and receiving an encrypted category vector generated by a classifier model, corresponding to the encryption key, from the encrypted transaction vector. The method further includes decrypting a category from the encrypted category vector with a decryption key corresponding to the encryption key and presenting the category.