Classification Model Learning Using Itemized Reliability Levels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques fail to accurately evaluate the reliability level of learning data when each item in the data set has a different reliability level, particularly in high expertise fields like medical image-based diagnosis, leading to inaccurate classification models.
Innovation Solution
An information processing apparatus and method that calculates itemized reliability levels for classification target data and labels, integrating these to determine the overall reliability of learning data, which is then used to formulate a classification model that adjusts the impact of varying data quality during machine learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If learning data from multiple creators with different expertise levels is used, then the quantity of learning data increases, but the reliability of individual learning data items varies and cannot be accurately evaluated
Solution Approach 1:
The patent segments the reliability evaluation into two distinct components: item reliability (evaluating each learning data item individually based on creator expertise) and dataset reliability (evaluating the overall learning dataset). This segmentation allows accurate evaluation of individual items while handling large quantities of diverse data from multiple creators.
Solution Approach 2:
The patent applies local quality by assigning different reliability levels to individual learning data items based on the specific expertise level of each creator. Instead of treating all data uniformly, each data item receives a localized reliability assessment corresponding to its creator's qualifications, enabling differentiated handling of high-quality and low-quality individual items within the larger dataset.
2Ease of operation
If traditional reliability evaluation methods are used that consider only creator information, then the evaluation process is simple, but the reliability level of individual learning data cannot be evaluated accurately when expertise is required
Solution Approach 1:
The patent changes the evaluation parameters from simple creator identification to creator expertise level assessment. By introducing expertise level as a quantifiable parameter (e.g., beginner, intermediate, expert), the system maintains operational simplicity while achieving precise reliability evaluation for individual learning data items in specialized fields.
Data Source
AI summary
An information processing apparatus includes an itemized reliability level calculation unit configured to calculate a first reliability level, wherein the first reliability level is a reliability level of classification target data, and a second reliability level, wherein the second reliability level is a reliability level of a label associated with the classification target data, a learning data reliability level calculation unit configured to calculate a learning data reliability level of learning data including the classification target data and the label based on the first reliability level and the second reliability level, and a classification model learning unit configured to formulate a classification model for giving a label to desired classification target data based on plural pieces of learning data and learning data reliability levels.