Local AI Data Processing With Adaptive Neural Network Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computing modules for AI models deployed on local devices face challenges in processing large amounts of data efficiently, leading to resource constraints and reduced reliability.
Innovation Solution
A data processing method involving local deployment of neural network processors with a combination of residual and attention processing units, optimizing the number and type of processing units based on vector sizes and correlations to reduce resource consumption and enhance efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If computing modules for AI models are deployed on local devices, then data processing reliability is improved, but resource consumption increases and efficiency decreases
Solution Approach 1:
The patent segments the computing module into multiple processing units (first processing units and second processing units) that operate in parallel. Each processing unit handles specific feature vector transformations, dividing the large-scale computation into smaller, manageable tasks that can be executed efficiently on local device hardware, thereby maintaining reliability while improving overall processing throughput.
Solution Approach 2:
The patent dynamically adjusts the processing pipeline based on vector sizes. Different processing units are activated depending on whether the input feature vectors have large or small dimensions, allowing the system to optimize resource allocation in real-time. This dynamic adaptation enables efficient local processing by matching computational resources to the actual data characteristics.
2Productivity
If the number of processing units is increased to handle large vector sizes, then processing capability is improved, but device complexity and resource consumption increase
Solution Approach 1:
The patent applies different processing strategies to different parts of the computation based on vector size characteristics. First processing units are dedicated to handling large vector sizes with specific optimization, while second processing units handle smaller vectors. This localized optimization approach allows the system to achieve high processing capability for each vector type without requiring a uniformly complex system architecture.
Solution Approach 2:
The system changes processing parameters (number and type of processing units) based on the input vector size parameter. When large vectors are detected, the system activates first processing units with appropriate configurations; when small vectors are present, second processing units are used. This parameter-based adaptation allows the system to maintain high productivity while controlling device complexity by only activating necessary processing units.
Data Source
AI summary
A data processing method includes obtaining input data, and using a local data processing model of an electronic device to locally process the input data to obtain target data. The electronic device includes a neural network processor. The local data processing model runs on the neural network processor, and thereby the local data processing model expediates local processing of the input data.