Parameter Training Efficiency Aware Resource Allocation for AIGC in Space-Air-Ground Integrated Networks

2024-06-19

arXiv.org

AI Summary

This work addresses the challenge of customizing AIGC models for training in space-air-ground integrated networks under stringent wireless resource constraints. Method: We propose a training-efficiency-aware joint resource allocation framework centered on parameter-efficient fine-tuning (PEFT), jointly optimizing user association, data offloading, and communication-computation resource allocation. To tackle the inherent non-convex sum-of-ratios optimization, we first equivalently reformulate it into a tractable quadratically constrained quadratic program (QCQP) or semidefinite program (SDP), integrating graph-theoretic modeling and fractional programming to ensure convergence. Results: Experiments demonstrate that our framework achieves an average 32.7% improvement in parameter training efficiency per unit resource over baseline methods, while consistently converging to high-quality stationary points—significantly enhancing both the feasibility and effectiveness of edge-based intelligent model training.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

With the evolution of artificial intelligence-generated content (AIGC) techniques and the development of space-air-ground integrated networks (SAGIN), there will be a growing opportunity to enhance more users' mobile experience with customized AIGC applications. This is made possible through the use of parameter-efficient fine-tuning (PEFT) training alongside mobile edge computing. In this paper, we formulate the optimization problem of maximizing the parameter training efficiency of the SAGIN system over wireless networks under limited resource constraints. We propose the Parameter training efficiency Aware Resource Allocation (PARA) technique to jointly optimize user association, data offloading, and communication and computational resource allocation. Solid proofs are presented to solve this difficult sum of ratios problem based on quadratically constrained quadratic programming (QCQP), semidefinite programming (SDP), graph theory, and fractional programming (FP) techniques. Our proposed PARA technique is effective in finding a stationary point of this non-convex problem. The simulation results demonstrate that the proposed PARA method outperforms other baselines.

Problem

• Integrative Network
• Resource Allocation
• Personalized Training

Innovation

• PARA Technology
• Mobile Edge Computing
• Parameter Efficient Fine-tuning