Hierarchical Feature Extraction for Robust Pattern Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern recognition methods face challenges in achieving robust identification while minimizing processing costs and reducing the likelihood of identification errors, particularly in varying patterns.
Innovation Solution
A hierarchical feature extraction method that involves a first feature extraction step, analysis of the distribution of the extraction results, and a second feature extraction step based on the analyzed distribution, allowing for the extraction of higher-layer features to enhance pattern recognition accuracy and reduce processing costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the number of types of features to be extracted is increased to increase robustness against pattern variance, then identification reliability is improved, but processing cost increases
Solution Approach 1:
The patent segments the feature extraction process into multiple hierarchical layers (first layer, second layer, and potentially more layers). Each layer extracts features at a different level of abstraction, allowing the system to achieve robust identification without extracting all possible features simultaneously. This segmented approach reduces processing cost while maintaining reliability.
Solution Approach 2:
The patent performs preliminary feature extraction at the first layer to obtain basic features before proceeding to higher layers. The distribution analysis of first-layer features is conducted in advance to determine which second-layer features to extract. This preliminary action avoids the need to extract all higher-order features regardless of necessity, thereby reducing processing cost while ensuring identification reliability.
2Productivity
If the number of types of features to be extracted is not increased, then processing cost is reduced, but the possibility of identification errors increases
Solution Approach 1:
The patent introduces a dynamic feature extraction process where the number and type of features extracted at each layer depend on the distribution analysis results from the previous layer. This dynamic approach ensures that only necessary features are extracted, reducing processing cost while maintaining identification accuracy by adapting the feature extraction to the specific characteristics of the input data.
Solution Approach 2:
The patent implements a feedback mechanism where the distribution of features extracted at one layer is analyzed and used to guide the extraction of features at the next layer. This feedback loop ensures that feature extraction is optimized based on actual data characteristics, preventing both over-extraction (which increases cost) and under-extraction (which decreases accuracy).
3Reliability
If hierarchical feature extraction is performed to achieve robust identification, then identification reliability is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by having different feature extraction processes at different hierarchical layers. Each layer has specialized extraction mechanisms tailored to the type of features appropriate for that level of abstraction. This localized specialization achieves robust identification without requiring the entire system to be uniformly complex.
Data Source
AI summary
Pattern recognition capable of robust identification for the variance of an input pattern is performed with a low processing cost while the possibility of identification errors is decreased. In a pattern recognition apparatus which identifies the pattern of input data from a data input unit (11) by using a hierarchical feature extraction processor (12) which hierarchically extracts features, an extraction result distribution analyzer (13) analyzes a distribution of at least one feature extraction result obtained by a primary feature extraction processor (121). On the basis of the analytical result, a secondary feature extraction processor (122) performs predetermined secondary feature extraction.