Classification Apparatus Using Attention Data for Feature Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing classification methods independently generate feature values for classification models, lacking a method to effectively integrate or weigh these values for improved classification accuracy.
Innovation Solution
A classification apparatus and method that acquire and process first and second types of feature values through intermediate feature extraction and attention data generation, allowing for the computation of enhanced feature values for classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple feature values are independently generated for classification models, then the classification process can handle different types of data, but the classification accuracy is limited due to lack of integration between feature values
Solution Approach 1:
The patent introduces attention data as an intermediary element that mediates between multiple independently generated feature values and the classification model. The attention data generation unit computes attention data from intermediate feature values, and this attention data is then used to weight and integrate the feature values before classification. This intermediary mechanism enables effective integration of multiple feature types without requiring complex direct interactions between them, thereby improving classification accuracy while managing processing complexity.
Solution Approach 2:
The patent transforms intermediate feature values into attention data through parameter changes in the form of attention weights. These attention weights dynamically adjust the importance of different feature values based on their relevance to the classification task. By changing the parameters (attention weights) rather than the fundamental structure of feature processing, the system achieves improved integration of multiple feature types without proportionally increasing system complexity.
2Measurement precision
If attention data is computed from intermediate feature values to weight feature importance, then classification accuracy improves through better feature integration, but computational complexity increases
Solution Approach 1:
The patent performs preliminary computation of intermediate feature values and attention data before the actual classification process. By pre-computing these elements in advance, the system prepares the weighted feature representations that will be used for classification, reducing the computational burden during the actual classification execution. This preliminary action allows the system to achieve better feature integration without proportionally increasing real-time computational power requirements.
Solution Approach 2:
The patent computes attention data for integrating multiple feature values, which represents a partial action toward full feature integration. Rather than exhaustively processing all possible feature combinations, the system selectively computes attention data from intermediate feature values to achieve sufficient integration for accurate classification. This partial action approach achieves the necessary level of feature integration without the excessive computational cost of complete feature space exploration.
Data Source
AI summary
A classification apparatus acquires first input data being a first type of feature value and second input data being a second type of feature value for a classification target. The classification apparatus computes a first intermediate feature value from first input data, and computes a second intermediate feature value from second input data. The classification apparatus computes first attention data from the second intermediate feature value, and computes second attention data from the first intermediate feature value. The classification apparatus computes a first feature value from the first intermediate feature value and the first attention data, and computes a second feature value from the second intermediate feature value and the second attention data. The classification apparatus performs classification for a classification target by using the first feature value, the second feature value, or both of the first feature value and the second feature value.