Information Processing System for Learning Data Duplication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In maintenance of image forming apparatuses, machine learning models struggle with reduced learning accuracy due to recognition of ground truth labels as different when replacement parts are of varying scales, leading to decreased data usage and accuracy, especially when replacement is performed on a part-by-part basis in some regions and unit-by-unit in others.
Innovation Solution
An information processing system that collects learning data by associating error history information with part replacement information, and duplicates replacement part information to align with unit-level data, creating second association data to enhance learning data accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data is collected from multiple regions with different replacement scales (part-level and unit-level), then the amount of learning data increases, but the learning accuracy decreases due to inconsistent ground truth labels
Solution Approach 1:
The invention segments the replacement part information into two distinct types: actual replacement parts (from error history) and candidate replacement parts (from part replacement information). By separating these data elements and processing them through different association methods, the system maintains label consistency while utilizing diverse data sources, thus resolving the contradiction between data quantity and learning accuracy
Solution Approach 2:
The invention creates duplicate association data with consistent ground truth labels by copying and standardizing replacement part information across different scales. Through the duplication unit, data from both part-level and unit-level replacements are transformed into a unified format with consistent labels, allowing the system to multiply usable data without compromising accuracy
2Measurement precision
If learning is performed using only data with the same scale, then the learning accuracy is maintained, but the amount of usable learning data decreases
Solution Approach 1:
The invention creates a universal data processing framework that handles both part-level and unit-level replacement data through the same association and duplication mechanisms. The system establishes universal association rules that work across different data scales, transforming heterogeneous data into a unified format that maintains consistency while expanding the usable data pool for machine learning
Data Source
AI summary
In an information processing system that collects learning data, an error history information acquiring unit configured to acquire error history information of a target device, a part replacement information acquisition unit configured to acquire part replacement information of the target device, an association unit configured to generate first association data in which the error history information and the part replacement information are associated with each other, and a duplication unit configured to generate second association data in which information on a predetermined replacement part is replaced with information on another replacement part that includes the predetermined replacement part and associated with the error history information.