Homomorphic Arithmetic Circuit for Private Decision Tree Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models, such as decision trees, face challenges in preserving privacy when evaluating sensitive data, as white-box or black-box access can lead to model inversion attacks compromising the privacy of training data.
Innovation Solution
A client-server protocol using homomorphic encryption allows for the evaluation of decision trees on encrypted data, where a plaintext dataset is homomorphically encrypted and transmitted to a server for evaluation using a homomorphic arithmetic circuit, ensuring that only the classification result is revealed to the client, while keeping the intermediate and final computational results private.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a decision tree model is used for classification, then classification accuracy is improved, but privacy of training data is compromised due to model inversion attacks
Solution Approach 1:
The patent introduces homomorphic encryption as an intermediary layer between the decision tree model and the input data. The encryption scheme allows the server to evaluate the decision tree on encrypted data without decrypting it, thereby preventing model inversion attacks while maintaining classification accuracy. The encrypted data acts as a mediator that preserves privacy during the evaluation process.
2Loss of information
If homomorphic encryption is applied to evaluate the decision tree, then privacy is preserved, but computational complexity increases
Solution Approach 1:
The patent segments the decision tree evaluation process into discrete arithmetic operations (additions and multiplications) that can be performed homomorphically on encrypted data. By breaking down the complex evaluation into smaller, manageable arithmetic steps, the system can leverage homomorphic encryption properties while controlling computational complexity through efficient circuit design.
3Ease of operation
If a standard Boolean comparison circuit is used for integer comparison, then comparison functionality is achieved, but the number of multiplications increases
Solution Approach 1:
The patent changes the approach to integer comparison by using an arithmetic circuit formulation instead of a standard Boolean comparison circuit. This parameter change in the computational model reduces the number of multiplications required for comparison operations, making the homomorphic evaluation more efficient while maintaining the comparison functionality needed for decision tree evaluation.
Data Source
AI summary
A non-interactive protocol is provided for evaluating machine learning models such as decision trees. A client can delegate the evaluation of a machine learning model such as a decision tree to a server by sending an encrypted input and receiving only the encryption of the result. The inputs can be encoded as vector of integers using their binary representation. The server can then evaluate the machine learning model using a homomorphic arithmetic circuit. The homomorphic arithmetic circuit provides an implementation that requires fewer multiplication than a Boolean comparison circuit. Efficient data representations are then combined with different algorithmic optimizations to keep the computational overhead and the communication cost low. Related apparatus, systems, techniques and articles are also described.