Edge-Cloud Model Adaptation for IoT Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning models in IoT environments face challenges in resource consumption and latency during data analysis and model updates, particularly in real-time anomaly detection, due to high computational requirements and inefficient distribution of processing tasks between edge and cloud computing devices.
Innovation Solution
A method for model adaptation is introduced, where a machine learning model is deployed on a first computing device for data analysis, with abnormal data portions transmitted to a second device with higher computing capability for updates, allowing the model to be redeployed and adapted over time, optimizing resource usage and reducing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If machine learning model training and updates are performed on edge devices with limited computing capability, then model adaptation can be achieved, but resource consumption and processing time increase significantly
Solution Approach 1:
The system segments the machine learning workflow into two distinct parts: (1) inference execution on edge devices with limited resources, and (2) model training and updates on cloud servers with abundant computing power. This segmentation allows each component to operate in its optimal environment, resolving the contradiction between model adaptation capability and resource consumption.
Solution Approach 2:
The patent introduces an intermediary mechanism where the edge device sends only necessary data samples or gradients to the cloud for model updates, rather than transferring all raw data. This intermediary approach enables model adaptation while minimizing resource consumption for data transmission and processing.
2Adaptability or versatility
If machine learning model training and updates are performed on edge devices, then model adaptation can be achieved, but processing speed and latency increase
Solution Approach 1:
By segmenting the computational workload, the patent enables real-time inference on edge devices while delegating time-consuming training operations to cloud servers. This resolves the latency issue as the edge device does not wait for lengthy training processes during inference operations.
Solution Approach 2:
The system performs model training and updates in advance on cloud servers, so that when the edge device needs to perform inference, the model is already optimized and ready. This preliminary action on the cloud side eliminates waiting time during critical inference operations on the edge.
3Measurement precision
If all data is transmitted to cloud devices for model updates, then model accuracy can be improved, but data transmission resource consumption increases
Solution Approach 1:
The patent extracts only the essential elements needed for model updates (such as selected data samples, features, or gradient information) rather than transmitting all raw data. This extraction approach maintains model accuracy while dramatically reducing data transmission volume and associated energy consumption.
Solution Approach 2:
The system employs partial action by transmitting only a subset of data that is most valuable for model updates, rather than all available data. This partial transmission strategy achieves sufficient model accuracy improvement while minimizing transmission resource consumption.
Data Source
AI summary
Embodiments of the present disclosure provide methods, devices, and computer program products for model adaptation. The method for model adaptation comprises: receiving, at a first computing device, a data set to be analyzed from a data collector and determining abnormality of the data set to be analyzed using a machine learning model deployed at the first computing device. The method further comprises transmitting, based on the determined abnormality of the data set, at least a portion of data in the data set to a second computing device, for update of the machine learning model, the second computing device having a higher computing capability than the first computing device. The method further comprises obtaining redeployment of the updated machine learning model from the second computing device.