Edge Device Machine Learning Model Update via Centralized Server
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Edge devices face limitations in processing and storing large volumes of data due to limited resources, which hampers the implementation of machine learning algorithms, leading to inefficient data analysis and potential biases from isolated learning.
Innovation Solution
A system where a centralized server trains a global machine learning model using data from multiple edge devices, updating local models on each device to improve performance and reduce biases by sharing insights across the network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If edge devices store and process large volumes of raw data locally, then data analysis flexibility and completeness improve, but device resource limitations (storage, computing power) are exceeded
Solution Approach 1:
The system divides the machine learning workflow into two segments: data collection and preprocessing at the edge device, and model training at the centralized server. This segmentation allows edge devices to operate within their resource constraints while still contributing to flexible and comprehensive data analysis through the centralized model training process.
Solution Approach 2:
A centralized server acts as an intermediary between multiple edge devices, collecting data from them and performing centralized model training. This intermediary handles the computationally intensive tasks that would overwhelm individual edge devices, enabling them to maintain resource efficiency while achieving sophisticated data analysis capabilities.
2Measurement precision
If edge devices train machine learning models using only local data, then operational independence is maintained, but model accuracy and generalization deteriorate due to local biases and anomalies
Solution Approach 1:
The system merges data from multiple edge devices to train a centralized machine learning model. By combining diverse data sources, the model achieves better generalization and reduced bias compared to training on single-device data, while still allowing edge devices to operate independently using the shared model.
Solution Approach 2:
The system implements a feedback mechanism where the centralized server trains models using aggregated data from multiple edge devices, then distributes these improved models back to the edge devices. This feedback loop continuously enhances prediction accuracy and model generalization while maintaining operational independence at the edge.
3Measurement precision
If edge devices communicate frequently with the centralized server, then model performance improves through continuous updates, but network dependency increases and operational efficiency decreases when disconnected
Solution Approach 1:
The system performs preliminary model training at the centralized server using aggregated data from multiple edge devices, then distributes the trained models to edge devices in advance. This preliminary action allows edge devices to operate independently with high-performance models even when disconnected from the server, reducing network dependency while maintaining model quality.
Data Source
AI summary
Disclosed is a technique that can be performed by an electronic device. The electronic device can generate time-stamped events, extract training data from the time-stamped events, and send the training data over a network to a remote computer. The electronic device can receive model data generated by the remote computer from the training data by use of a machine learning process, update a local model of the electronic device based on the received model data, and generate an output by processing locally sourced data of the electronic device with the updated local model.