Automated ML Model Selection for Network Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In virtualized data centers, analyzing and troubleshooting network operations is complex due to the lack of uniformity in machine learning (ML) models and the need for ad-hoc additions, which increases complexity and storage requirements, making it difficult for administrators to determine the appropriate ML model for specific insights.
Innovation Solution
A network analysis system that collects data from multiple network devices, trains multiple ML models, and automatically selects the best model based on evaluation metrics to generate predictions, reducing the need for ad-hoc additions and simplifying the process of obtaining insights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple ML models are maintained and ad-hoc additions are made to provide different network insights, then the system's versatility and adaptability improve, but the device complexity and storage requirements increase
Solution Approach 1:
The patent implements a unified ML system that can perform multiple network analysis functions (traffic prediction, anomaly detection, performance optimization) through a single standardized model interface. The system uses common data structures and model formats that can be applied across different network scenarios, eliminating the need for separate specialized models for each insight type while maintaining versatility.
2Adaptability or versatility
If multiple ML models are maintained and ad-hoc additions are made to provide different network insights, then the system's versatility and adaptability improve, but the storage requirements increase
Solution Approach 1:
The patent implements a unified ML system that can perform multiple network analysis functions (traffic prediction, anomaly detection, performance optimization) through a single standardized model interface. The system uses common data structures and model formats that can be applied across different network scenarios, eliminating the need for separate specialized models for each insight type while maintaining versatility.
3Ease of operation
If administrators manually select appropriate ML models for specific insights, then the ease of operation improves, but the time and effort required increases
Solution Approach 1:
The patent implements automated model selection mechanisms where the system autonomously determines the appropriate ML model based on the specific network insight required. The standardized model interface and automated selection process eliminate the need for administrator intervention in model choice, reducing both the complexity of operation and the time to obtain predictions while maintaining ease of use.
Data Source
AI summary
This disclosure describes techniques that include using an automatically trained machine learning system to generate a prediction. In one example, this disclosure describes a method comprising: based on a request for the prediction: training each respective machine learning (ML) model in a plurality of ML models to generate a respective training-phase prediction in a plurality of training-phase predictions; automatically determining a selected ML model in the plurality of ML models based on evaluation metrics for the plurality of ML; and applying the selected ML model to generate the prediction based on data collected from a network that includes a plurality of network devices.