Mixture of Experts for Intelligent Networks: A Large Language Model-enabled Approach

被引:0
|
作者
Du, Hongyang [1 ]
Liu, Guangyuan [1 ]
Lin, Yijing [2 ]
Niyato, Dusit [1 ]
Kang, Jiawen [3 ,4 ,5 ]
Xiong, Zehui [6 ]
Kim, Dong In [7 ]
机构
[1] Nanyang Technol Univ, Sch Comp Sci & Engn, Singapore 639798, Singapore
[2] Beijing Univ Posts & Telecommun, State Key Lab Networking & Switching Technol, Beijing 100876, Peoples R China
[3] Guangdong Univ Technol, Sch Automat, Guangzhou 510006, Peoples R China
[4] Minist Educ, Key Lab Intelligent Informat Proc & Syst Integrat, Guangzhou 510006, Peoples R China
[5] Guangdong HongKong Macao Joint Lab Smart Discrete, Guangzhou 510006, Peoples R China
[6] Singapore Univ Technol & Design, Pillar Informat Syst Technol & Design, Singapore 487372, Singapore
[7] Sungkyunkwan Univ, Dept Elect & Comp Engn, Suwon 16419, South Korea
来源
20TH INTERNATIONAL WIRELESS COMMUNICATIONS & MOBILE COMPUTING CONFERENCE, IWCMC 2024 | 2024年
基金
中国国家自然科学基金; 新加坡国家研究基金会;
关键词
Generative AI (GAI); large language model; mixture of experts; network optimization;
D O I
10.1109/IWCMC61514.2024.10592370
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Optimizing various wireless user tasks poses a significant challenge for networking systems because of the expanding range of user requirements. Despite advancements in Deep Reinforcement Learning (DRL), the need for customized optimization tasks for individual users complicates developing and applying numerous DRL models, leading to substantial computation resource and energy consumption and can lead to inconsistent outcomes. To address this issue, we propose a novel approach utilizing a Mixture of Experts (MoE) framework, augmented with Large Language Models (LLMs), to analyze user objectives and constraints effectively, select specialized DRL experts, and weigh each decision from the participating experts. Specifically, we develop a gate network to oversee the expert models, allowing a collective of experts to tackle a wide array of new tasks. Furthermore, we innovatively substitute the traditional gate network with an LLM, leveraging its advanced reasoning capabilities to manage expert model selection for joint decisions. Our proposed method reduces the need to train new DRL models for each unique optimization problem, decreasing energy consumption and AI model implementation costs. The LLM-enabled MoE approach is validated through a general maze navigation task and a specific network service provider utility maximization task, demonstrating its effectiveness and practical applicability in optimizing complex networking systems.
引用
收藏
页码:531 / 536
页数:6
相关论文
共 49 条
  • [1] Human-Autonomy Teaming on Autonomous Vehicles with Large Language Model-Enabled Human Digital Twins
    Cui, Can
    Ma, Yunsheng
    Cao, Xu
    Ye, Wenqian
    Wang, Ziran
    2023 IEEE/ACM SYMPOSIUM ON EDGE COMPUTING, SEC 2023, 2023, : 319 - 324
  • [2] Virtual Co-Pilot: Multimodal Large Language Model-enabled Quick-access Procedures for Single Pilot Operations
    Li, Fan
    Feng, Shanshan
    Yan, Yuqi
    Lee, Ching-Hung
    Ong, Yew Soon
    2024 IEEE CONFERENCE ON ARTIFICIAL INTELLIGENCE, CAI 2024, 2024, : 1501 - 1506
  • [3] CECA: An intelligent large-language-model-enabled method for accounting embodied carbon in buildings
    Gu, Xierong
    Chen, Cheng
    Fang, Yuan
    Mahabir, Ron
    Fan, Lei
    BUILDING AND ENVIRONMENT, 2025, 272
  • [4] Overcoming language barriers via machine translation with sparse Mixture-of-Experts fusion of large language models
    Zhu, Shaolin
    Jian, Dong
    Xiong, Deyi
    INFORMATION PROCESSING & MANAGEMENT, 2025, 62 (03)
  • [5] Efficient Inference Offloading for Mixture-of-Experts Large Language Models in Internet of Medical Things
    Yuan, Xiaoming
    Kong, Weixuan
    Luo, Zhenyu
    Xu, Minrui
    ELECTRONICS, 2024, 13 (11)
  • [6] Large Language Model Firewall for AIGC Protection with Intelligent Detection Policy
    Huang, Tianrui
    You, Lina
    Cai, Nishui
    Huang, Ting
    2024 2ND INTERNATIONAL CONFERENCE ON MOBILE INTERNET, CLOUD COMPUTING AND INFORMATION SECURITY, MICCIS 2024, 2024, : 247 - 252
  • [7] Artificially Intelligent Billing in Spine Surgery: An Analysis of a Large Language Model
    Zaidat, Bashar
    Lahoti, Yash S.
    Yu, Alexander
    Mohamed, Kareem S.
    Cho, Samuel K.
    Kim, Jun S.
    GLOBAL SPINE JOURNAL, 2025, 15 (02) : 1113 - 1120
  • [8] GeoLLM: A specialized large language model framework for intelligent geotechnical design
    Xu, Hao-Ruo
    Zhang, Ning
    Yin, Zhen-Yu
    Njock, Pierre Guy Atangana
    COMPUTERS AND GEOTECHNICS, 2025, 177
  • [9] Large Language Model Agents Enabled Generative Design of Fluidic Computation Interfaces
    Lu, Qiuyu
    Fang, Jiawei
    Yao, Zhihao
    Yang, Yue
    Lyu, Shiqing
    Mi, Haipeng
    Yao, Lining
    PROCEEDINGS OF THE 37TH ANNUAL ACM SYMPOSIUM ON USER INTERFACE SOFTWARE AND TECHNOLOGY, UIST ADJUNCT 2024, 2024,
  • [10] Adversarial Text Purification: A Large Language Model Approach for Defense
    Moraffah, Raha
    Khandelwal, Shubh
    Bhattacharjee, Amrita
    Liu, Huan
    ADVANCES IN KNOWLEDGE DISCOVERY AND DATA MINING, PT V, PAKDD 2024, 2024, 14649 : 65 - 77