Transformed Data Training for Privacy-Preserving Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Companies face challenges in deriving aggregated insights from user data due to user reluctance in sharing private information, making it difficult to create effective classifiers without compromising data privacy.
Innovation Solution
A computer-implemented method for determining multiple training samples by randomly selecting and combining data samples based on their labels, allowing for the creation of classifier coefficients without disclosing actual data, ensuring differential privacy and maintaining data security.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If actual user data is shared with third parties for classifier training, then classification accuracy and model performance are improved, but user privacy and data security are compromised
Solution Approach 1:
The patent segments the training process into two distinct phases: (1) local processing where each user device creates transformed training samples from their own data without sharing raw data, and (2) centralized aggregation where the server collects these transformed samples for model training. This segmentation allows the server to receive training data without accessing actual user information, resolving the contradiction between training quality and privacy protection.
Solution Approach 2:
The patent introduces transformed training samples as an intermediary representation between raw user data and the classification model. These transformed samples contain sufficient information for effective model training while being mathematically designed to prevent reconstruction of original user data. This intermediary form enables knowledge transfer without direct data sharing, addressing the privacy-accuracy tradeoff.
2Object-affected harmful factors
If transformed training samples are created by combining multiple data samples, then data privacy is protected through aggregation, but the complexity of the training sample generation process increases
Solution Approach 1:
The patent implements self-service by enabling each user device to autonomously generate its own transformed training samples using local computational resources. The transformation process is performed independently at the edge device without requiring complex centralized processing or coordination with other users. This distributes the computational burden and simplifies the overall system architecture while maintaining strong privacy guarantees.
Data Source
AI summary
This disclosure relates to learning with transformed data such as determining multiple training samples from multiple data samples. Each of the multiple data samples comprises one or more feature values and a label that classifies that data sample. A processor determines each of the multiple training samples by randomly selecting a subset of the multiple data samples, and combining the feature values of the data samples of the subset based on the label of each of the data samples of the subset. Since the training samples are combinations of randomly chosen data samples, the training samples can be provided to third parties without disclosing the actual training data. This is an advantage over existing methods in cases where the data is confidential and should therefore not be shared with a learner of a classifier, for example.