Adversarial Training: Enhancing Out-of-Distribution Generalization for Learning Wireless Resource Allocation

2025/06/26 by Liu, Shengjie, Yang, Chenyang
#FOS: Electrical engineering #Signal Processing (eess.SP) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2506.21208

Abstract

Deep neural networks (DNNs) have widespread applications for optimizing resource allocation. Yet, their performance is vulnerable to distribution shifts between training and test data, say wireless channels. In this paper, we resort to adversarial training (AT) for enhancing out-of-distribution (OOD) generalizability of DNNs trained in unsupervised manner. We reformulate AT problem to reflect the OOD degradation, and propose a one-step gradient ascent algorithm to solve the AT problem for training DNNs. The proposed method is evaluated by optimizing hybrid precoding. Simulation results showcase the enhanced OOD performance of multiple kinds of DNNs, with approximately 5\(∼\)20% improvement, across various channel distributions, even when the samples only from a single distribution (e.g., Rayleigh fading) are used for training.

Citations

• Achieving Robust Channel Estimation Neural Networks by Designed Training Data
• Recursive GNNs for Learning Precoding Policies with Size-Generalizability
• Data Augmentation of Bridging the Delay Gap for DL-based Massive MIMO CSI Feedback
• Domain Generalization in Machine Learning Models for Wireless Communications: Concepts, State-of-the-Art, and Open Issues
• Achieving Robust Generalization for Wireless Channel Estimation Neural Networks by Designed Training Data
• Towards Out-Of-Distribution Generalization: A Survey
• Improved OOD Generalization via Adversarial Training and Pre-training
• Robust Reconfigurable Intelligent Surfaces via Invariant Risk and Causal\n Representations
• Adversarial Machine Learning in Wireless Communications using RF Data: A Review
• A Family of Deep Learning Architectures for Channel Estimation and Hybrid Beamforming in Multi-Carrier mm-Wave Massive MIMO
• Learning to Optimize with Unsupervised Learning: Training Deep Neural Networks for URLLC
• Towards Deep Learning Models Resistant to Adversarial Attacks
• Distributionally Robust Stochastic Optimization with Wasserstein Distance
• Distributionally Robust Stochastic Optimization with Wasserstein Distance

Related

• Improving Learning Efficiency for Wireless Resource Allocation with Symmetric Prior
• Constructing Deep Neural Networks with a Priori Knowledge of Wireless Tasks
• Structure of Deep Neural Networks with a Priori Information in Wireless Tasks
• Distilling Knowledge from Resource Management Algorithms to Neural Networks: A Unified Training Assistance Approach
• AirNet: Neural Network Transmission over the Air