Homomorphic Fraud Model Computation on Encrypted Transaction Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Financial institutions face challenges in maintaining the confidentiality of predictive fraud models and transaction data while sharing them with third parties due to regulatory restrictions, and existing methods risk exposing sensitive information.

Innovation Solution

Implementing homomorphic encryption schemes that allow performing verifiable computations on encrypted data, enabling financial institutions to apply privately trained predictive fraud models to encrypted third-party transaction data while maintaining confidentiality by using homomorphic public and private keys.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If financial institutions share predictive fraud models and transaction data with third parties, then fraud detection accuracy is improved, but data confidentiality and model security are compromised

Engineering Contradiction:

Improvefraud detection accuracyVSAvoiddata confidentiality

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

Homomorphic encryption acts as an intermediary that enables third-party fraud detection models to process encrypted transaction data without accessing the actual data or model coefficients. The encryption scheme allows computations to be performed on ciphertext, producing encrypted results that can be decrypted without revealing intermediate values, thus maintaining confidentiality while enabling accurate fraud detection.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent transforms the data from plaintext to ciphertext using homomorphic encryption, changing the state of the data while preserving its computational properties. This parameter change allows the data to be processed in encrypted form, maintaining security while enabling fraud detection functionality.

Inventive Principle:

Principle #35Parameter changes

2Loss of information

If financial institutions use traditional encryption methods to protect transaction data, then data confidentiality is maintained, but third parties cannot perform fraud detection computations

Engineering Contradiction:

Improvedata confidentialityVSAvoidcomputational capability on encrypted data

Core Design Contradiction:

Loss of informationVSAdaptability or versatility

Solution Approach 1:

Homomorphic encryption serves as an intermediary that bridges the gap between data confidentiality and computational accessibility. It allows third-party fraud detection models to perform computations on encrypted data without requiring decryption, enabling both security and computational versatility simultaneously.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent changes the encryption approach from traditional symmetric or asymmetric encryption to homomorphic encryption, which preserves the ability to perform computations on encrypted data. This parameter change in the encryption scheme enables third parties to process encrypted transaction data while maintaining confidentiality.

Inventive Principle:

Principle #35Parameter changes

3Reliability

If financial institutions share model coefficients with third parties, then fraud detection accuracy is improved, but model security and intellectual property are compromised

Engineering Contradiction:

Improvefraud detection accuracyVSAvoidmodel security

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

Homomorphic encryption acts as an intermediary that enables third-party fraud detection models to operate on encrypted data without accessing the actual model coefficients. The encryption scheme allows the models to be applied to encrypted transaction data, producing accurate fraud detection results while keeping the model coefficients confidential.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Loss of information

If financial institutions encrypt transaction data before sharing with third parties, then data confidentiality is maintained, but existing encryption methods do not allow computations on the encrypted data

Engineering Contradiction:

Improvedata confidentialityVSAvoidcomputational operation on encrypted data

Core Design Contradiction:

Loss of informationVSEase of operation

Solution Approach 1:

The patent changes the encryption methodology from traditional encryption to homomorphic encryption, which preserves computational capabilities. This parameter change allows third parties to perform fraud detection computations on encrypted transaction data without requiring decryption, maintaining both confidentiality and operational ease.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12512960B2Homomorphic computations on encrypted data within a distributed computing environment

Publication Date: 2025.12.30 THE TORONTO DOMINION BANK

AI summary

The disclosed exemplary embodiments include computer-implemented systems, apparatuses, and processes that perform homomorphic computations on encrypted third-party data within a distributed computing environment. For example, an apparatus receives a homomorphic public key and encrypted transaction data characterizing an exchange of data from a computing system, and encrypts modelling data associated with a first predictive model, such as a machine learning model or an artificial neural network model, using the homomorphic public key. The apparatus may perform homomorphic computations that apply the first predictive model to the encrypted transaction data in accordance with the encrypted first modelling data, and transmit an encrypted first output of the homomorphic computations to the computing system, which may decrypt the encrypted first output using a homomorphic private key and generate decrypted output data indicative of a predicted likelihood that the data exchange represents fraudulent activity.