Encrypted Risk Scoring With Homomorphic Inner Products
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for computing risk scores and risk categorization on sensitive data require decryption, which exposes the data to untrusted parties and are inefficient, especially when large amounts of data and complex computations are involved, rendering them impractical for real-time applications.
Innovation Solution
A method for computing risk scores and categorization using encrypted data through inner product operations and encryption key processes, allowing computations to be performed directly on encrypted data without decryption, using techniques like homomorphic encryption to maintain data privacy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If standard encryption is used to protect secret data, then data privacy is improved, but computational capability deteriorates because computations cannot be performed on encrypted data
Solution Approach 1:
The patent introduces an intermediary encryption scheme that enables computations on encrypted data by using a trapdoor function mechanism. The encryption scheme includes a public key for encoding data and a private key for decoding results, allowing an untrusted party to perform computations on ciphertexts without accessing the plaintext. This intermediary cryptographic structure resolves the contradiction by enabling both privacy protection and computational capability simultaneously.
2Ease of operation
If trusted hardware is used to perform computations on decrypted data, then computational capability is improved, but security deteriorates because data must be exposed to untrusted parties
Solution Approach 1:
The patent inverts the conventional approach by performing computations on encrypted data rather than decrypting data first. The encryption scheme is designed with a trapdoor function where the public key enables encoding and computation, while the private key enables decoding. This inversion allows computations to be performed in the encrypted domain, eliminating the need to expose sensitive data to untrusted hardware while maintaining full computational capability.
3Reliability
If homomorphic encryption is used to compute on encrypted data, then data privacy is improved, but computational speed deteriorates due to the complexity of operations on encrypted data
Solution Approach 1:
The patent changes the parameters of the encryption scheme by using a specific trapdoor function-based encryption that allows efficient computation on ciphertexts. The encryption uses a public key for encoding that enables straightforward computational operations, and the private key for decoding that efficiently recovers the result. This parameter change in the cryptographic approach maintains data privacy while significantly improving computational speed compared to traditional homomorphic encryption methods.
Data Source
AI summary
Methods and system for risk determination and risk categorization using encrypted data are provided. The risk determination can involve determining an inner product operation between a generalized weight table and an encrypted incidence vector, summing the result of the inner product operation and/or decrypting the results. Method and systems for encrypting data for use in homomorphic risk determination are also provided.