Edge Device Anomaly Detection via Federated Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing transaction processing systems face challenges in detecting anomalous user behavior due to privacy measures implemented by operating systems and browsers, which restrict the transmission of private user data for training machine learning models.
Innovation Solution
A hybrid approach to anomaly detection is implemented, where machine learning models are trained partially or fully on edge devices (federated learning) rather than a central server, ensuring private user data remains on the device and reducing the need for data transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If user data is transmitted to a central server for training machine learning models, then the system can detect anomalous user behavior effectively, but privacy measures implemented by operating systems and browsers restrict this data transmission
Solution Approach 1:
Instead of transmitting user data from edge devices to a central server for training machine learning models, the patent inverts the approach by bringing the training process to the edge devices themselves. Each device trains local machine learning models using its own user data, eliminating the need for data transmission while maintaining anomaly detection capability.
Solution Approach 2:
The patent introduces an intermediary mechanism where a central server provides initial model configurations and receives trained local models from edge devices, but does not directly handle user data. This intermediary architecture enables collaborative learning without requiring sensitive data transmission between devices and servers.
2Loss of information
If a hybrid federated learning approach is implemented to maintain user privacy, then data transmission is reduced, but the complexity of the system architecture increases
Solution Approach 1:
The patent segments the machine learning training process into independent local training units at each edge device, with each device autonomously training its own model using local data. This segmentation distributes the computational complexity across multiple devices rather than concentrating it in a central server, managing overall system complexity while protecting user privacy.
Solution Approach 2:
The patent creates a universal framework where the same federated learning architecture and model training process can be applied across diverse edge devices and applications. This multi-functional approach uses standardized protocols and model structures that work universally across different devices, reducing the complexity burden despite the distributed architecture.
Data Source
AI summary
Techniques are disclosed in which an edge user computing device pre-processes a stream of user data prior to using the stream of data to train a machine learning model at the edge device. The edge device receives the stream of user data, where the stream of data includes a first set of characteristics associated with the edge device and a second set of characteristics associated with a plurality of user requests received from a user of the edge device. The edge device repeatedly generates, using the stream of data, sets of pre-processed user data by performing pre-processing techniques on characteristics included in the stream of data. The edge device repeatedly trains, using the sets of pre-processed data, a baseline model to generate a device-trained model, where the baseline model is trained at the edge device without providing user data included in the stream of data to a server computer system.