Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Explanation-Guided Fair Federated Learning for Transparent 6G RAN Slicing

Swastika Roy, Hatim Chergui, Christos Verikoukis

TL;DR

An explanation-guided federated learning (EGFL) scheme is designed to ensure trustworthy predictions by exploiting the model explanation emanating from XAI strategies during the training run time via Jensen-Shannon (JS) divergence to measure and validate the faithfulness of the explanations quantitatively.

Abstract

Future zero-touch artificial intelligence (AI)-driven 6G network automation requires building trust in the AI black boxes via explainable artificial intelligence (XAI), where it is expected that AI faithfulness would be a quantifiable service-level agreement (SLA) metric along with telecommunications key performance indicators (KPIs). This entails exploiting the XAI outputs to generate transparent and unbiased deep neural networks (DNNs). Motivated by closed-loop (CL) automation and explanation-guided learning (EGL), we design an explanation-guided federated learning (EGFL) scheme to ensure trustworthy predictions by exploiting the model explanation emanating from XAI strategies during the training run time via Jensen-Shannon (JS) divergence. Specifically, we predict per-slice RAN dropped traffic probability to exemplify the proposed concept while respecting fairness goals formulated in terms of the recall metric which is included as a constraint in the optimization task. Finally, the comprehensiveness score is adopted to measure and validate the faithfulness of the explanations quantitatively. Simulation results show that the proposed EGFL-JS scheme has achieved more than $50\%$ increase in terms of comprehensiveness compared to different baselines from the literature, especially the variant EGFL-KL that is based on the Kullback-Leibler Divergence. It has also improved the recall score with more than $25\%$ relatively to unconstrained-EGFL.

Explanation-Guided Fair Federated Learning for Transparent 6G RAN Slicing

TL;DR

An explanation-guided federated learning (EGFL) scheme is designed to ensure trustworthy predictions by exploiting the model explanation emanating from XAI strategies during the training run time via Jensen-Shannon (JS) divergence to measure and validate the faithfulness of the explanations quantitatively.

Abstract

Future zero-touch artificial intelligence (AI)-driven 6G network automation requires building trust in the AI black boxes via explainable artificial intelligence (XAI), where it is expected that AI faithfulness would be a quantifiable service-level agreement (SLA) metric along with telecommunications key performance indicators (KPIs). This entails exploiting the XAI outputs to generate transparent and unbiased deep neural networks (DNNs). Motivated by closed-loop (CL) automation and explanation-guided learning (EGL), we design an explanation-guided federated learning (EGFL) scheme to ensure trustworthy predictions by exploiting the model explanation emanating from XAI strategies during the training run time via Jensen-Shannon (JS) divergence. Specifically, we predict per-slice RAN dropped traffic probability to exemplify the proposed concept while respecting fairness goals formulated in terms of the recall metric which is included as a constraint in the optimization task. Finally, the comprehensiveness score is adopted to measure and validate the faithfulness of the explanations quantitatively. Simulation results show that the proposed EGFL-JS scheme has achieved more than increase in terms of comprehensiveness compared to different baselines from the literature, especially the variant EGFL-KL that is based on the Kullback-Leibler Divergence. It has also improved the recall score with more than relatively to unconstrained-EGFL.
Paper Structure (17 sections, 1 theorem, 26 equations, 8 figures, 3 tables, 1 algorithm)

This paper contains 17 sections, 1 theorem, 26 equations, 8 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Explanation Guided Learning Concept
  4. RAN Architecture and Datasets
  5. Explanation-Guided FL for Transparent Traffic Drop Prediction
  6. Closed-Loop Description
  7. Model Testing and Explanation
  8. Feature Masking
  9. Divergence Module
  10. Explanation-Guided Fair Traffic Drop Prediction
  11. EGFL Convergence Analysis
  12. Results
  13. Parameter Settings and Baseline
  14. Result Analysis
  15. Navigating EGFL: Methodological Insights and XAI Integration Significance
  16. ...and 2 more sections

Key Result

Theorem 1

We consider that EGFL follows a geometric failure model and fails to meet the recall constraint with an average violation rate $0<\nu<1$. It is also assumed that an oracle $\mathcal{O}_{\delta}$ with error---$\delta$ excluding the JS divergence---optimizes $\mathcal{L}_{\mathbf{W}{k,n}^{(t)}}$, whil where and $\alpha = \delta+\ln(1-V^2/4)$ while $V$ stands for the total variation distance between

Figures (8)

  • Figure 1: Federated traffic drop probability prediction in 6G RAN NS.
  • Figure 2: Explanation Guided FL building blocks
  • Figure 3: Analysis of FL training loss vs FL rounds of EGFL with lower bound of recall score, $\gamma = [0.82, 0.85, 0.84]$
  • Figure 4: Analysis of recall score with lower bound $\gamma = [0.82, 0.85, 0.84]$
  • Figure 5: Analysis of comprehensiveness score with lower bound of recall score, $\gamma = [0.82, 0.85, 0.84]$
  • ...and 3 more figures

Theorems & Definitions (3)

  • Definition 1: Approximate Bayesian Optimization Oracle
  • Theorem 1: EGFL Convergence Analysis
  • proof