Tree-Based Data Sanitization for ML Security and Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for ensuring the absence of criticality during data processing, such as data sanitization, expert determination, and externalizing critical operations, are costly, complex, and affect the quality of output, particularly in machine learning and AI model training, due to over-sanitization, frequent expert determinations, and added costs and complexities.
Innovation Solution
A computer-implemented method that sanitizes data processing operations by analyzing tree-based structures to identify and transform data signals associated with criticality conditions, applying self, up, and root transforms to avoid creating criticality in intermediate operations, allowing for flexible and high-accuracy data processing without omitting or altering data unnecessarily.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data sanitization is applied to ensure absence of criticality, then security and privacy are improved, but data quality deteriorates due to over-sanitization
Solution Approach 1:
The patent segments the data processing operation into multiple executable plans with tree-based structures. Each plan is independently analyzed for criticality conditions, allowing selective sanitization only where necessary rather than blanket sanitization of all data, thus preserving data quality while maintaining security.
Solution Approach 2:
The system applies sanitization locally to specific data signals and plans that are determined to create criticality conditions, rather than applying uniform sanitization across all data. This localized approach preserves data quality in non-critical areas while ensuring security in critical areas.
2Reliability
If expert determination is used to ascertain absence of criticality, then reliability is improved, but productivity deteriorates due to frequent expert determinations required when data changes
Solution Approach 1:
The system implements automated criticality analysis that evaluates data signals and plans without requiring external expert determination. The automated system independently determines whether criticality conditions exist and applies sanitization accordingly, eliminating the need for frequent expert reviews when data changes, thus maintaining reliability while improving productivity.
3Reliability
If critical operations are externalized to separate entities, then security is improved, but device complexity increases
Solution Approach 1:
The patent merges the criticality analysis and sanitization functions within the same data processing system. The system includes components that automatically analyze plans for criticality conditions and apply sanitization transformations internally, eliminating the need to externalize critical operations to separate entities, thus maintaining security while reducing system complexity.
4Reliability
If data sanitization transforms data to avoid criticality, then security is improved, but manufacturing precision deteriorates due to unnecessary transformations
Solution Approach 1:
The system applies sanitization partially and selectively only to data signals and plans that are determined to create criticality conditions. By avoiding excessive sanitization of non-critical data, the system maintains security while preserving data processing accuracy and minimizing unnecessary transformations.
Data Source
AI summary
Flexible, high-accuracy data processing techniques and accompanying systems avoid criticality in intermediate computations through intelligent, low cost sanitization of data operations. A data processing operation including one or more plans is received, with each plan having a data operation described as a tree based-structure. The plans that are determined to create criticality on execution are sanitized by transforming the plan itself, ancestors, and/or children of the plan. Determining whether execution of a plan creates criticality is based on the determination of whether a set of criticality conditions includes data signals that are associated with the plan. After sanitization, the data processing operation can be fully executed without criticality arising in intermediate operations.