Information Processing Apparatus Confidential Data Detection Learned Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multifunction peripherals using machine learning risk confidential information leakage through learned models, as these models can inadvertently expose sensitive data when copied or used across different domains.
Innovation Solution
An information processing apparatus and method that determines whether learned models stored in its system contain confidential information and presents this information to users, allowing for secure management and access control to prevent leakage, including features like access authority settings and model reset options.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If machine learning is used to improve user convenience through learned models, then user convenience and functionality are improved, but confidential information may leak through the learned models
Solution Approach 1:
The system performs preliminary detection to determine whether learned models include confidential information before the models are used or copied. This advance detection prevents confidential information leakage by identifying at-risk models before they can be compromised, while still allowing safe models to be freely used for improving user convenience.
Solution Approach 2:
The patent introduces an intermediary detection mechanism that acts as a mediator between the learned models and potential copying operations. This intermediary layer detects and prevents the transmission of confidential information without interfering with the legitimate use of non-confidential models, thus maintaining user convenience while blocking harmful information flow.
2Productivity
If learned models are copied and used across different domains, then model utility and productivity are improved, but confidential information security deteriorates
Solution Approach 1:
Before allowing learned models to be copied or distributed across different domains, the system performs preliminary detection to identify models containing confidential information. This pre-check ensures that only safe models are copied for productive use, while models with confidential information are protected from unauthorized distribution.
Solution Approach 2:
The system implements a feedback mechanism that provides information about the detection results to users and system administrators. This feedback loop allows organizations to understand which models contain confidential information and take appropriate actions, thereby maintaining both model utility and information security across different domains.
Data Source
AI summary
An information processing apparatus and a method of controlling the information processing apparatus are provided. The information processing apparatus stores a plurality of learned models, determines whether the stored plurality of learned models include confidential information, and presents, to a user, learned models of the plurality of learned models determined to include the confidential information.