Learning Unit Adjusts Data Contribution for Classification Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information processing systems for class classification in series data face inefficiencies in learning processes, as they do not effectively adjust the contribution of each data set based on the ease of classification, leading to suboptimal learning outcomes.
Innovation Solution
An information processing system that includes an acquisition unit for obtaining series data elements, a calculation unit for determining a likelihood ratio, a classification unit for classifying data, and a learning unit that adjusts the contribution of each data set based on its classification ease, thereby optimizing the learning process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If uniform learning contribution is applied to all series data, then the learning process is simple to implement, but the classification accuracy for hard-to-classify data deteriorates
Solution Approach 1:
The patent applies local quality by assigning different learning contribution degrees to different series data based on their individual classification difficulty. The learning contribution degree is determined locally for each data point according to its specific characteristics (e.g., likelihood ratio, classification confidence), rather than applying a uniform contribution degree to all data. This allows the learning process to focus more on hard-to-classify data while maintaining simplicity in the overall framework.
2Productivity
If all series data are given equal weight in learning, then the learning process is computationally efficient, but the classification performance on difficult data deteriorates
Solution Approach 1:
The patent implements dynamics by making the learning contribution degree variable rather than static. The contribution degree dynamically adjusts based on the classification difficulty of each series data, which is determined through metrics such as likelihood ratios or classification confidence scores. This dynamic adjustment allows the system to automatically allocate more learning resources to difficult data points while maintaining computational efficiency through automated weighting.
3Measurement precision
If the learning process focuses on hard-to-classify data, then classification accuracy improves, but the complexity of determining data importance increases
Solution Approach 1:
The patent applies feedback by using the classification results and confidence metrics as feedback signals to determine the learning contribution degree. The system evaluates how well each series data is classified (through likelihood ratios or confidence scores) and uses this feedback to adjust the learning contribution accordingly. This feedback mechanism automatically identifies hard-to-classify data without requiring complex manual evaluation, as the classification performance itself provides the necessary information for weighting.
Data Source
AI summary
An information processing system includes: an acquisition unit that obtains a plurality of elements included in series data; a calculation unit that calculates a likelihood ratio indicating a likelihood of a class to which the series data belong, on the basis of at least two consecutive elements of the plurality of elements: a classification unit that classifies the series data into at least one class, on the basis of the likelihood ratio; and a learning unit that performs learning related to calculation of the likelihood ratio, by using a plurality of series data. The learning unit changes a degree of contribution to the learning of each of the plurality of series data in accordance with ease of classification of the series data. According to such an information processing system, it is possible to properly perform the learning related to the calculation of the likelihood ratio.