Scholarlink

<p>The controllability of complex temporal networks depends on identifying the minimum set of driver nodes (MDS), an NP-hard problem that is computationally prohibitive for large-scale systems. This paper proposes the TACO-BPNN framework, a novel hybrid approach that synergistically integrates a Temporal Ant Colony Optimization (TACO) algorithm with a predictive Backpropagation Neural Network (BPNN) to overcome this computational barrier. In our framework, the BPNN learns temporal features from the network to predict high-potential driver nodes, which effectively prunes the solution space and guides the TACO algorithm’s metaheuristic search toward the optimal MDS. Our evaluations on four real-world temporal networks demonstrate the framework’s performance and effectiveness: TACO-BPNN reduces the required driver nodes by over 30% compared to a standalone TACO and over 60% compared to a Genetic Algorithm, while simultaneously cutting computation time by more than 25%. Ultimately, this research presents a potent hybrid intelligence paradigm for network control, offering a scalable and effective pathway to managing the dynamics of complex temporal systems.</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

TACO-BPNN: a hybrid ant colony optimization and backpropagation neural network approach for efficient controllability of temporal networks

  • 期刊论文

  • 发表日期：

    2026年1月8日

查看全文
  • Peyman Arebi,
  • Sina Keshvadi

摘要

The controllability of complex temporal networks depends on identifying the minimum set of driver nodes (MDS), an NP-hard problem that is computationally prohibitive for large-scale systems. This paper proposes the TACO-BPNN framework, a novel hybrid approach that synergistically integrates a Temporal Ant Colony Optimization (TACO) algorithm with a predictive Backpropagation Neural Network (BPNN) to overcome this computational barrier. In our framework, the BPNN learns temporal features from the network to predict high-potential driver nodes, which effectively prunes the solution space and guides the TACO algorithm’s metaheuristic search toward the optimal MDS. Our evaluations on four real-world temporal networks demonstrate the framework’s performance and effectiveness: TACO-BPNN reduces the required driver nodes by over 30% compared to a standalone TACO and over 60% compared to a Genetic Algorithm, while simultaneously cutting computation time by more than 25%. Ultimately, this research presents a potent hybrid intelligence paradigm for network control, offering a scalable and effective pathway to managing the dynamics of complex temporal systems.