Neural Network Sampling for Pattern Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current artificial neural networks face challenges in increasing the recognition rate of specific patterns due to limitations in feature map extraction and pattern recognition methods.
Innovation Solution
The proposed solution involves generating a first and second format image from an input image using a format converter, and sampling these images using different sampling schemes to create first and second feature maps, which are then used to operate the artificial neural network, thereby enhancing pattern recognition capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single sampling scheme is used to generate feature maps, then the device complexity is low, but the recognition rate of specific patterns is insufficient
Solution Approach 1:
The sampling process is segmented into multiple independent sampling schemes (first sampling scheme and second sampling scheme), each generating separate feature maps. This segmentation allows the system to capture different pattern characteristics through diverse sampling approaches, thereby improving recognition rate while maintaining manageable complexity through modular processing
Solution Approach 2:
The patent introduces an additional dimension by applying multiple sampling schemes instead of a single scheme. This dimensional expansion in the sampling space enables the extraction of complementary feature information from the same input image, enhancing pattern recognition capability without proportionally increasing device complexity
2Measurement precision
If multiple feature maps are generated using different sampling schemes, then the pattern recognition performance is enhanced, but the processing time increases
Solution Approach 1:
The format converter performs preliminary action by converting the input image into multiple format images before the sampling stage. This pre-processing ensures that when multiple sampling schemes are applied, the data is already in the appropriate format, reducing the overall processing time despite generating multiple feature maps
Solution Approach 2:
The learning unit merges the results from multiple feature maps generated by different sampling schemes. By combining these complementary feature representations, the system achieves enhanced pattern recognition performance while the merging process itself is optimized to minimize additional processing time
Data Source
AI summary
An apparatus for an artificial neural network includes a format converter, a sampling unit, and a learning unit. The format converter generates a first format image and a second format image based on an input image. The sampling unit samples the first format image using a first sampling scheme to generate a first feature map, and samples the second format image using a second sampling scheme different from the first sampling scheme to generate a second feature map. The learning unit operates the artificial neural network using the first feature map and the second feature map.