Convolutional Neural Network Feature Extraction via Cone of Dependency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Convolutional neural networks require significant memory and computational resources for feature extraction in image recognition and classification tasks, leading to performance bottlenecks due to large data sets and memory access inefficiencies.
Innovation Solution
Implementing a cone of dependency and cone of influence based processing method, combined with local cache memory and partitioning of images into sections, to reduce main memory accesses and optimize feature extraction processing by keeping only necessary data in local cache for efficient calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If convolutional neural networks process large high-resolution image data sets, then image recognition and classification accuracy is improved, but memory access costs and computational resource requirements increase significantly
Solution Approach 1:
The patent divides the image data set into multiple smaller sections or chunks that can be processed independently. This segmentation allows the system to load and process only the necessary portions of data into local cache memory, reducing overall memory access costs while maintaining the ability to analyze complete images for accurate recognition.
Solution Approach 2:
The patent implements different processing qualities for different parts of the data. Critical regions or frequently accessed data are kept in local cache memory for fast processing, while less critical data are processed from main memory. This local quality differentiation optimizes the balance between processing speed and memory access efficiency.
2Measurement precision
If convolutional neural networks process large high-resolution image data sets, then image recognition and classification accuracy is improved, but computational processing time increases
Solution Approach 1:
By segmenting the image data into smaller sections, the system can process them in parallel or in smaller batches, reducing the time required to load and process the entire data set. This segmentation enables more efficient utilization of computational resources and reduces overall processing time while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary processing of image data by pre-loading necessary data into local cache memory and pre-processing sections of images before they are needed for classification. This preliminary action reduces the time required during the actual recognition and classification operations.
3Use of energy by moving object
If convolutional neural networks use local cache memory for feature extraction, then memory access efficiency is improved, but device complexity increases
Solution Approach 1:
The patent introduces local cache memory as an intermediary between the main memory and the computational units. This intermediary layer simplifies the overall system architecture by providing a straightforward buffer for data storage and retrieval, making the memory hierarchy more manageable despite the added complexity of multi-level memory access.
4Productivity
If convolutional neural networks partition images into sections for processing, then computational efficiency is improved, but data processing completeness may be compromised
Solution Approach 1:
The patent segments images into sections that are processed independently, but ensures that the segmentation does not compromise completeness by using overlapping regions or by processing multiple segments in a coordinated manner. This approach maintains processing completeness while improving computational efficiency through parallel processing.
Solution Approach 2:
The system incorporates feedback mechanisms that monitor and verify the processing of image sections, ensuring that all necessary information is captured and processed correctly. This feedback loop maintains reliability and processing completeness while allowing the benefits of section-based processing to be realized.
Data Source
AI summary
Artificial intelligence is an increasingly important sector of the computer industry. One of the most important applications for artificial intelligence is object recognition and classification from digital images. Convolutional neural networks have proven to be a very effective tool for object recognition and classification from digital images. However, convolutional neural networks are extremely computationally intensive thus requiring high-performance processors, significant computation time, and significant energy consumption. To reduce the computation time and energy consumption a “cone of dependency” and “cone of influence” processing techniques are disclosed. These two techniques arrange the computations required in a manner that minimizes memory accesses such that computations may be performed in local cache memory. These techniques significantly reduce the time to perform the computations and the energy consumed by the hardware implementing a convolutional neural network.