Learning precoding policies with neural networks enables low complexity online implementation, robustness to channel impairments, and joint optimization with channel acquisition. However, existing neural networks suffer from high training complexity and poor generalization ability when they are used to learn to optimize precoding for mitigating multi-user interference. This impedes their use in practical systems where the number of users is time-varying. In this paper, we propose a graph neural network (GNN) to learn precoding policies by harnessing both the mathematical model and the property of the policies. We first show that a vanilla GNN cannot well-learn pseudo-inverse of channel matrix when the numbers of antennas and users are large, and is not generalizable to unseen numbers of users. Then, we design a GNN by resorting to the Taylor's expansion of matrix pseudo-inverse, which allows for capturing the importance of the neighbored edges to be aggregated that is crucial for learning precoding policies efficiently. Simulation results show that the proposed GNN can well learn spectral efficient and energy efficient precoding policies in single- and multi-cell multi-user multi-antenna systems with low training complexity, and can be well generalized to the numbers of users.
Snowflake: Scaling GNNs to High-Dimensional Continuous Control via Parameter Freezing
University of Oxford(牛津大学)
最近的研究表明,图神经网络(GNN)可以学习与典型的多层感知器(MLP)一样有效的运动控制策略,具有卓越的传输和多任务性能(Wang et al.,2018; Huang等人,2020年)。到目前为止,结果仅限于对小型代理的训练,随着传感器和执行器数量的增加,GNN的性能迅速恶化。在监督学习环境中使用GNN的一个关键动机是它们对大型图的适用性,但这种好处尚未实现运动控制。我们确定了导致这种不良伸缩性的常见GNN架构的弱点:网络中编码,解码和传播消息的MLP中的过拟合。为了解决这个问题,我们引入了Snowflake,这是一种用于高维连续控制的GNN训练方法,可以冻结网络中遭受过拟合的部分参数。Snowflake显著提高了GNN在大型代理上的运动控制性能,现在与MLP的性能相匹配,并且具有优异的传输特性。
Efficient GNN Training Through Structure-Aware Randomized Mini-Batching
NVIDIA USA(NVIDIA美国分公司);NVIDIA & Stanford University USA(NVIDIA与斯坦福大学);NVIDIA & Bar-Ilan University Israel(NVIDIA与巴伊兰大学);NVIDIA & Technion University Israel(NVIDIA与技术学院)
RouteNet-Erlang: A Graph Neural Network for Network Performance Evaluation
Barcelona Neural Networking Center(巴塞罗那神经网络中心);Universitat Politècnica de Catalunya(加泰罗尼亚理工大学);AGH University of Science and Technology(AGH科技大学);Huawei Technologies Co., Ltd.(华为技术有限公司)
Fast State-Augmented Learning for Wireless Resource Allocation with Dual Variable Regression
Department of Electrical and Systems Engineering, University of Pennsylvania(宾夕法尼亚大学电气与系统工程系);Department of Biostatistics and Bioinformatics, Duke University(杜克大学生物统计与生物信息学系);Johns Hopkins Applied Research Laboratory (APL)(约翰霍普金斯应用研究实验室)
GPEvac: GNN-Based PPO for Adaptive Evacuation Routing During Shooting Events
The Bredesen Center for Interdisciplinary Research and Graduate Education(布雷德森跨学科研究与研究生教育中心);Department of Mechanical and Aerospace Engineering, University of Tennessee(田纳西大学机械与航空航天工程系)
Graph Neural Network based Agent in Google Research Football
Central South University of Forestry and Technology(中南林业科技大学);Fordham University(福特汉姆大学);Zhejiang University(浙江大学)
深度神经网络(DNN)可以近似价值函数或策略用于强化学习,这使得强化学习算法更强大。然而,在强化学习的特定情况下,一些DNN,如卷积神经网络(CNN),无法提取足够的信息或需要太长时间才能从输入中获得足够的特征。例如,Google Research Football的输入数据是球员位置的小地图,这是一个训练代理踢足球的强化学习环境。这些信息不仅包含在玩家的坐标中,还包含在不同玩家之间的关系中。CNN既不能提取足够的信息,也不需要太长的训练时间。为了解决这个问题,本文提出了一种以图神经网络(GNN)为模型的深度q学习网络(DQN)。GNN将输入数据转换为更好地表示足球运动员位置的图,以便提取不同运动员之间交互的更多信息。通过两个GNN来近似其本地和目标价值函数,该DQN允许玩家通过使用价值函数来了解每个预期动作的预期价值,从而从他们的经验中学习。所提出的模型证明了GNN在足球比赛中的强大功能,其表现优于其他DRL模型,步骤明显减少。
Ultra-Reliable Risk-Aggregated Sum Rate Maximization via Model-Aided Deep Learning
Department of Electrical and Computer Engineering, Yale University, New Haven, USA(耶鲁大学电子与计算机工程系);Department of Mathematics, King’s College London, London, England, UK(伦敦国王学院数学系)
Limit Analysis of Graph Neural Networks with Wireless Conflict Graphs
Department of Electrical and Systems Engineering, University of Pennsylvania(宾夕法尼亚大学电气与系统工程系);Halıcıoğlu Data Science Institute, UCSD(加州大学圣地亚哥分校Halıcıoğlu数据科学研究所)
FLEXIBLE: Forecasting Cellular Traffic by Leveraging Explicit Inductive Graph-Based Learning
Orange Innovation(奥兰治创新公司);Nantes University(南特大学);École Centrale Nantes(南特中央理工学院);IMT Atlantique(大西洋电信与技术学院);CNRS(法国国家科学研究中心);INRIA(法国国家信息与自动化研究所);LS2N(南特数字科学实验室);University of Luxembourg(卢森堡大学);SnT(卢森堡大学安全、可靠性和信任中心)
Adaptive Soft Error Protection for Neural Network Processing
Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences(中国科学院大学杭州高等研究院);State Key Lab of Processors, Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所国家处理器实验室);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);Jinan Inspur Data Technology Co., Ltd.(济南Inspur数据技术有限公司)