Device-Specific ANN Models for Distributed Edge Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data analytics platforms face challenges in efficiently processing large and varied datasets using neural networks, leading to computational intensity and difficulty in wider adoption and accuracy.
Innovation Solution
A distributed data analytics framework that generates device-specific artificial neural network (ANN) models for edge devices like smartphones and IoT devices, optimizing model execution and training using device-specific parameters and environmental conditions, and employs multiple ANN models to enhance accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex neural network models are used for data analytics, then processing accuracy is improved, but computational intensity and execution difficulty increase
Solution Approach 1:
The patent segments the neural network execution by generating device-specific models tailored to individual edge devices. Each device receives a customized model optimized for its hardware capabilities, dividing the complex processing task into device-appropriate computational units rather than forcing a single monolithic model on all devices.
Solution Approach 2:
The system implements local quality by creating device-specific neural network models that are optimized for each individual device's hardware characteristics. Each device receives models tailored to its specific processor, memory, and computational capabilities, ensuring optimal performance for that local configuration rather than using a generic model.
2Adaptability or versatility
If device-specific neural network models are generated for each user device, then model accuracy and adaptability are improved, but system complexity and processing time increase
Solution Approach 1:
The system performs preliminary action by pre-generating device-specific neural network models before actual data processing occurs. Models are created in advance based on device specifications and training data, so that when data needs processing, the optimized models are already ready for immediate execution without adding complexity during the processing task itself.
Solution Approach 2:
The patent uses copying by creating customized versions of neural network models for each device based on a base model architecture. Rather than creating entirely new models from scratch, the system copies and adapts a foundational model structure to fit each device's specific requirements, reducing overall system complexity while maintaining device-specific optimization.
3Measurement precision
If multiple neural network models are deployed across different devices, then processing accuracy is improved, but data transmission and model distribution requirements increase
Solution Approach 1:
The system extracts only the essential model parameters and architecture information that are needed for device-specific customization. By separating the core model logic from device-specific optimizations, the patent reduces the amount of data that needs to be transmitted and stored while maintaining the ability to generate accurate device-specific models.
Data Source
AI summary
The invention provides systems and method for generating device-specific artificial neural network (ANN) models for distribution across user devices. Sample datasets are collected from devices in a particular environment or use case and include predictions by device-specific ANN models executing the user devices. The received datasets are used with existing datasets and stored ANN models to generate updated device-specific ANN models from each of the stored instances of the device ANN models based on the training data.