Integrated Likelihood Ratios for Correlated Sequential Data Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sequential probability ratio test (SPRT) methods assume independent and identically distributed elements in sequential data, leading to insufficient accuracy due to neglecting the relationship between data elements.
Innovation Solution
An information processing device and method that calculates indicators for each element considering multiple elements, integrates these indicators to determine an integrated indicator, and classifies sequential data based on this integrated indicator, accounting for the relationships between elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If SPRT is used to classify sequential data assuming independent and identically distributed elements, then the classification process is simple and fast, but the classification accuracy is insufficient when data elements have relationships
Solution Approach 1:
The patent segments the classification task into two stages: first calculating individual likelihood ratios for each data element, then integrating them to determine the final classification. This segmentation allows the system to handle dependent data while maintaining computational efficiency by processing elements individually before aggregation.
Solution Approach 2:
The patent merges the likelihood ratios of multiple data elements into a single integrated indicator through multiplication. This combining approach allows the system to capture relationships between dependent elements while maintaining a unified classification decision framework, resolving the accuracy-complexity contradiction.
2Measurement precision
If SPRT assumes independent and identically distributed elements, then the mathematical formula is simple, but it cannot accurately classify data with strong correlations between elements
Solution Approach 1:
The patent incorporates feedback by using the integrated likelihood ratio to dynamically update the classification decision. The system continuously integrates likelihood ratios as new elements are processed, allowing the classification to adapt to the cumulative evidence while accounting for relationships between elements.
Solution Approach 2:
The patent creates a composite classification approach by combining individual likelihood ratios into an integrated indicator. This composite approach preserves information about element relationships while maintaining a unified classification framework, eliminating the loss of relational information.
Data Source
AI summary
Provided is an information processing device including: an acquisition unit that sequentially acquires a plurality of elements included in sequential data; a first calculation unit that calculates, for each of the plurality of elements, indicators each indicating which of a plurality of classes is appropriate for corresponding element to belong to, in consideration of two or more elements among the plurality of elements; a second calculation unit that calculates, by integrating the indicators of the plurality of elements, an integrated indicator indicating which of the plurality of classes is appropriate for the sequential data to belong to; and a classification unit that classifies the sequential data into one of the plurality of classes based on the integrated indicator.