Scholarlink

<p>Coalition Formation among multiple agents is a valuable approach for human-centric applications involving large-scale data and complex interactions. This NP-hard problem closely resembles the multi-dimensional knapsack challenge. This article proposes a Quantum-Inspired Elephant Herd Algorithm (QEHO) for addressing this combinatorial optimization challenge. QEHO uses qubits for particle positions and adaptive quantum rotation for fine-grained distance-sensitive solution updates. A hybrid attractor mechanism combining influence of both local and global best solutions governs the magnitude of quantum rotation. QEHO’s efficacy was validated extensively across benchmark problems and a real-world factory dataset. For the Jooken benchmark, QEHO successfully converged on all 100 hard instances with exceptionally low numerical deviation and fast execution speed, whereas the reference algorithm had failed in half. For multidimensional knapsacks from OR library, QEHO performed consistently well across diverse tightness ratios with near-perfect linear correlation to reference results. In additional simulations on a warehouse setup, QEHO effectively allocated resources (agents) to tasks (coalitions) within given constraints. Furthermore, QEHO successfully optimized task schedules for a real-world human-robot collaborative assembly line, maximizing the ergonomic relief score of human workers to mitigate chronic strain injuries, thereby demonstrating its potential for designing human centric man-machine systems. Key findings from parameter tuning experiments include: (i) Manhattan/Hamming distance are the most effective binary distance metric; (ii) the algorithm performs best when the local and global influences are equally balanced; (iii) the update mechanism is sensitive to the trade-off between the local and global pulls; (iv) adaptive weights are more suitable than fixed weights for navigating complex search spaces; (v) aggressive updates help in escaping local optima without destabilizing the search. Overall, QEHO holds strong potential in solving large-scale, resource-constrained, and human-centric discrete optimization challenges.</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

Quantum-Inspired Elephant Herd Optimization Algorithm for Coalition Formation in Human-Centric Applications

  • 期刊论文

  • 发表日期：

    2026年2月11日

查看全文
  • Rupali Mitra,
  • Romit S. Beed,
  • Tamal Chakraborty

摘要

Coalition Formation among multiple agents is a valuable approach for human-centric applications involving large-scale data and complex interactions. This NP-hard problem closely resembles the multi-dimensional knapsack challenge. This article proposes a Quantum-Inspired Elephant Herd Algorithm (QEHO) for addressing this combinatorial optimization challenge. QEHO uses qubits for particle positions and adaptive quantum rotation for fine-grained distance-sensitive solution updates. A hybrid attractor mechanism combining influence of both local and global best solutions governs the magnitude of quantum rotation. QEHO’s efficacy was validated extensively across benchmark problems and a real-world factory dataset. For the Jooken benchmark, QEHO successfully converged on all 100 hard instances with exceptionally low numerical deviation and fast execution speed, whereas the reference algorithm had failed in half. For multidimensional knapsacks from OR library, QEHO performed consistently well across diverse tightness ratios with near-perfect linear correlation to reference results. In additional simulations on a warehouse setup, QEHO effectively allocated resources (agents) to tasks (coalitions) within given constraints. Furthermore, QEHO successfully optimized task schedules for a real-world human-robot collaborative assembly line, maximizing the ergonomic relief score of human workers to mitigate chronic strain injuries, thereby demonstrating its potential for designing human centric man-machine systems. Key findings from parameter tuning experiments include: (i) Manhattan/Hamming distance are the most effective binary distance metric; (ii) the algorithm performs best when the local and global influences are equally balanced; (iii) the update mechanism is sensitive to the trade-off between the local and global pulls; (iv) adaptive weights are more suitable than fixed weights for navigating complex search spaces; (v) aggressive updates help in escaping local optima without destabilizing the search. Overall, QEHO holds strong potential in solving large-scale, resource-constrained, and human-centric discrete optimization challenges.