A reinforcement learning-assisted genetic programming algorithm for team formation problem considering person-job matching

Efficient team formation is crucial for successful completion of new projects in a company. To address the team formation problem considering person-job matching (TFP-PJM), a 0-1 integer programming model with intuitionistic fuzzy numbers is proposed that considers both job matching and team members...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Neurocomputing (Amsterdam) Ročník 650; s. 130917
Hlavní autoři: Guo, Yangyang, Wang, Hao, He, Lei, Pedrycz, Witold, Suganthan, Ponnuthurai Nagaratnam, Xing, Lining, Song, Yanjie
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier B.V 14.10.2025
Témata:
Ensemble population strategy
Genetic programming
Intuitionistic fuzzy numbers
Reinforcement learning
Surrogate model
Team formation
Surrogate model
Intuitionistic fuzzy numbers
Ensemble population strategy
Genetic programming
Team formation
Reinforcement learning
ISSN:0925-2312
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Efficient team formation is crucial for successful completion of new projects in a company. To address the team formation problem considering person-job matching (TFP-PJM), a 0-1 integer programming model with intuitionistic fuzzy numbers is proposed that considers both job matching and team members’ willingness to communicate for improved efficiency. A reinforcement learning-assisted genetic programming (RL-GP) algorithm is introduced as it can flexibly combine heuristic rules to solve complex TFP-PJMs. Before each generation’s population search, the agent selects from four population search modes based on information obtained, named ensemble population strategy, to balance exploration and exploitation. Furthermore, a k-Nearest Neighbor-based surrogate model is used to evaluate individual-generated formation plans, speeding up the algorithm learning process. Comparison experiments demonstrate RL-GP’s overall performance and effectiveness of improved strategies within the algorithm. The hyper-heuristic rules obtained through efficient learning can be utilized as decision-making aids when forming project teams. This study reveals the advantages of reinforcement learning methods, ensemble strategies, and the surrogate model applied to the GP framework. The diversity and intelligent selection of search patterns, along with fast adaptation evaluation, are distinct features that enable RL-GP to be deployed in real-world enterprise environments.
ISSN:0925-2312
DOI:10.1016/j.neucom.2025.130917