Biometric Data Anonymization via Landmark Graph Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for anonymizing biometric data, such as k-anonymity, are inadequate for protecting biometric data like facial and fingerprint data, as they do not effectively maintain data utility while ensuring anonymity.
Innovation Solution
A global partitioning-based and random sampling-based method that uses weighted averages of landmarks and machine learning models to anonymize biometric data, representing it as a graph structure, ensuring k-anonymity and t-closeness while retaining the ability to detect attributes like emotions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing k-anonymity methods are used to anonymize biometric data, then anonymity is achieved, but data utility is lost
Solution Approach 1:
The patent transforms biometric data into a different parameter space by converting facial images into landmark graphs with specific topological features. This parameter transformation allows the data to maintain its essential characteristics for emotion detection while being rendered unusable for identity recognition, thus achieving both anonymity and utility preservation
Solution Approach 2:
The patent selectively modifies specific local features of the biometric data - specifically the topological relationships between facial landmarks - while preserving other local qualities such as the relative positions and distances that are important for emotion detection. This localized modification approach maintains data utility for intended purposes while eliminating identity information
2Object-affected harmful factors
If biometric data is anonymized to protect privacy, then individual identification is prevented, but the ability to detect attributes like emotions is reduced
Solution Approach 1:
The patent segments biometric data into distinct functional components: identity-related features (which are removed through anonymization) and attribute-related features (which are preserved). By separating these functions at the data structure level, the system can protect privacy while maintaining the capability to detect attributes like emotions with high precision
3Reliability
If traditional anonymization methods are applied to biometric datasets, then anonymity is enforced, but the dataset becomes less useful for downstream machine learning tasks
Solution Approach 1:
The patent introduces an intermediary transformation process that converts biometric data into a graph representation with specific topological properties. This intermediary form acts as a mediator between the original biometric data and downstream machine learning tasks, enabling anonymity while preserving the structural information needed for emotion detection and other attribute-based analyses
Data Source
AI summary
A global partitioning-based method for anonymizing a dataset of biometric data may include an anonymization computer program: (1) receiving a value k representing a number of records to hide a biometric datum among, a value t that represents a t-closeness parameter for a t-close distribution, a weight parameter, and a first number of features to retain for determining an attribute of interest; (2) receiving the attribute of interest; (3) calculating a distribution of the attribute of interest in a biometric dataset; (4) splitting the biometric dataset into a plurality of k-sized clusters that satisfy the t-close distribution; (5) anonymizing each biometric datum in the plurality of k-sized clusters using a weighted average of landmarks for the biometric datums in k-sized clusters using the weight parameter; (6) adding each anonymized biometric datum into an anonymized biometric dataset; and (7) persisting the anonymized biometric dataset.