Scholarlink

<p>Metaheuristics are broadly applicable problem-solving algorithms that typically operate without domain-specific knowledge but often fail to fully exploit the data they generate, resulting in suboptimal performance. This underutilization limits their ability to adapt behavior, tune parameters, and select effective search strategies, ultimately affecting convergence speed and solution quality. To address these limitations, this paper integrates multi-agent Deep Reinforcement Learning (RL) into Grey Wolf Optimizer (GWO) for the Influence Maximization (IM) task in social networks. The proposed algorithm, RLSetGWO, combines the learning capabilities of RL with the problem-solving power of metaheuristics, achieving significantly better solution quality than traditional metaheuristics with only a minor increase in runtime.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Using Reinforcement Learning to Boost Grey Wolf Optimizer for Influence Maximization in Social Networks

  • 期刊论文

  • 发表日期：

    2026年2月6日

查看全文
  • Mehdy Roayaei

摘要

Metaheuristics are broadly applicable problem-solving algorithms that typically operate without domain-specific knowledge but often fail to fully exploit the data they generate, resulting in suboptimal performance. This underutilization limits their ability to adapt behavior, tune parameters, and select effective search strategies, ultimately affecting convergence speed and solution quality. To address these limitations, this paper integrates multi-agent Deep Reinforcement Learning (RL) into Grey Wolf Optimizer (GWO) for the Influence Maximization (IM) task in social networks. The proposed algorithm, RLSetGWO, combines the learning capabilities of RL with the problem-solving power of metaheuristics, achieving significantly better solution quality than traditional metaheuristics with only a minor increase in runtime.