18065064. A LIGHTWEIGHT ATTACKER IDENTIFICATION METHOD FOR FEDERATED LEARNING WITH SECURE BYZANTINE-ROBUST AGGREGATION VIA CLUSTERING simplified abstract (Dell Products L.P.)
A LIGHTWEIGHT ATTACKER IDENTIFICATION METHOD FOR FEDERATED LEARNING WITH SECURE BYZANTINE-ROBUST AGGREGATION VIA CLUSTERING
Organization Name
Inventor(s)
Paulo Abelha Ferreira of Rio de Janeiro (BR)
Pablo Nascimento Da Silva of Niterói (BR)
Maira Beatriz Hernandez Moran of Rio de Janeiro (BR)
A LIGHTWEIGHT ATTACKER IDENTIFICATION METHOD FOR FEDERATED LEARNING WITH SECURE BYZANTINE-ROBUST AGGREGATION VIA CLUSTERING - A simplified explanation of the abstract
This abstract first appeared for US patent application 18065064 titled 'A LIGHTWEIGHT ATTACKER IDENTIFICATION METHOD FOR FEDERATED LEARNING WITH SECURE BYZANTINE-ROBUST AGGREGATION VIA CLUSTERING
Simplified Explanation: The patent application describes a method for detecting outliers in a cluster of nodes in a machine learning model by aggregating gradients and running a robust aggregation operation to calculate an outlier score.
- Central node receives gradients from each node in the cluster
- Gradients contain updates to a global machine learning model
- Aggregates gradients to obtain an aggregated gradient for the cluster
- Runs a robust aggregation operation on the aggregated gradients to calculate an outlier score
- Identifies the cluster as an outlier if the outlier score meets specified criteria
- Identifies suspicious nodes within the outlier cluster
Potential Applications: This technology can be applied in anomaly detection systems, fraud detection in financial transactions, and identifying outliers in large datasets.
Problems Solved: The technology addresses the challenge of detecting outliers in clusters of nodes in machine learning models efficiently and accurately.
Benefits: - Improved outlier detection in machine learning models - Enhanced security in fraud detection systems - Increased accuracy in anomaly detection processes
Commercial Applications: Potential commercial applications include fraud detection software, anomaly detection tools for cybersecurity, and outlier detection systems for large-scale data analysis.
Prior Art: Prior research in outlier detection methods in machine learning models and anomaly detection systems can provide insights into similar approaches and techniques.
Frequently Updated Research: Stay informed about advancements in outlier detection algorithms, robust aggregation methods, and anomaly detection techniques to enhance the efficiency and accuracy of the technology.
Questions about Outlier Detection Technology: 1. How does the method of aggregating gradients contribute to outlier detection in machine learning models? 2. What are the key factors that determine the outlier score for a cluster of nodes in the proposed technology?
Original Abstract Submitted
One method includes receiving, at a central node, a respective gradient from each node of a cluster, and each of the gradients comprises a respective update to a global machine learning model maintained at the central node, aggregating, by the central node, the gradients to obtain an aggregated gradient for the cluster, and the aggregated gradient is part of a list of aggregated gradients, running, by the central node, a robust aggregation operation on the aggregated gradients in the list to obtain an outlier score for the cluster, and when the outlier score equals or exceeds a specified value, or falls within a specified range of values, identifying the cluster as an outlier, and identifying nodes within the cluster as suspicious.
