Trustworthy and Explainable Federated System for Extracting Descriptive Rules in a Data Streaming Environment

María Asunción Padilla-Rascón, Ángel Miguel García-Vico, Cristóbal J. Carmona

Keywords: Federated Rule Learning, Evolutive Algorithm, Trustworthy Artificial Intelligence, Data Streaming, Supervised Descriptive Rules

TL;DR: A federated learning system using evolutionary algorithms to extract explainable and secure descriptive rules from streaming data while preserving privacy.

Abstract:

In the information age, continuous streams of data from connected devices require intelligent models that ensure security, privacy and transparency. Federated learning enables knowledge sharing while adhering to the principles of trustworthy AI. This work synthesizes the Trustworthy and Explainable Federated System for Extracting Descriptive Rules in a Data Streaming Environment (TEFeS-SDR) [1] study, which introduces an evolutionary single-objective federated system for extracting descriptive rules while prioritizing privacy and security through advanced encryption techniques (binary, symmetric, and asymmetric). It ensures traceability and transparency, and experimental results confirm its resilience to concept drift while maintaining high quality models, advancing responsible AI by integrating explainability, security and efficiency.

Cite:
[1] María Asunción Padilla Rascón, Ángel Miguel García-Vico, and Cristóbal J. Carmona. Trustworthy and explainable federated system for extracting descriptive rules in a data streaming environment. Results in Engineering, 25:104137, 2025.

Submission Number: 4