Unsupervised Labeling for Edge Device Transitional Intervals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In edge computing environments, there is a challenge in providing decision support for detecting anomalous events in mobile edge devices due to the lack of labels for divergent behavior, which is costly and impractical, especially with large and diverse datasets.
Innovation Solution
A system that collects and evaluates edge device trajectories using autoencoders to identify transitional intervals, generating labels for anomalous behavior through unsupervised methods, and using these labels to inform decision support systems for remedial actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If unsupervised labeling is used for edge device behavior, then labeling cost and time are reduced, but reproducibility and traceability of the labeling process deteriorate
Solution Approach 1:
The system implements feedback mechanisms where labeled transitional intervals are fed back into the clustering process to refine boundary definitions. This allows the unsupervised system to progressively improve its labeling accuracy and consistency, thereby enhancing governability while maintaining automated labeling efficiency.
Solution Approach 2:
The system performs preliminary actions by pre-defining cluster boundaries and transitional interval frameworks before actual labeling occurs. This preparation enables systematic and reproducible labeling processes, addressing the governability issue while maintaining the time efficiency of unsupervised learning.
2Measurement precision
If manual labeling is used for edge device behavior, then labeling precision is improved, but cost and practicality deteriorate
Solution Approach 1:
The system enables self-service labeling where the unsupervised learning model automatically identifies and labels transitional intervals without human intervention. The algorithm uses clustering techniques to autonomously define normal behavior patterns and detect anomalies, achieving both precision and practicality.
Solution Approach 2:
The system changes parameters such as clustering thresholds and boundary definitions to optimize labeling precision. By dynamically adjusting these parameters based on data characteristics, the system maintains high accuracy while eliminating the need for expensive manual labeling processes.
3Reliability
If large diverse datasets are used for edge device behavior, then model accuracy is improved, but data processing complexity and cost increase
Solution Approach 1:
The system extracts only the critical information needed for labeling by identifying and isolating transitional intervals from the large diverse datasets. This extraction process reduces data processing complexity while maintaining model accuracy, as the system focuses computational resources on the most informative portions of the data.
Solution Approach 2:
The system segments the complex data processing task into distinct stages: data collection, clustering, transitional interval identification, and labeling. This segmentation allows the system to handle large diverse datasets systematically, reducing overall complexity while improving model accuracy through focused processing of relevant patterns.
Data Source
AI summary
One example method includes collecting trajectories from each edge device in a group of edge devices, extracting a respective latent space vector from each trajectory, evaluating the latent space vectors to identify a respective set of one or more transitional intervals corresponding to each of the latent space vectors, and labeling the transitional intervals. The trajectories include information about the movement of the edge devices within a domain, and the labels on the transitional intervals indicate whether, and to what extent, the movement of the edge devices conforms to expected movements.