Reconstructing Networks from Profit Sequences in Evolutionary Games via a Multiobjective Optimization Approach with...

Area of Science:

• Complexity Science
• Network Science
• Computational Game Theory

Background:

• Evolutionary games (EG) model interactions in complex natural and social systems.
• Reconstructing EG networks is vital for understanding system dynamics but challenging with existing methods.
• Current methods like lasso require difficult-to-tune parameters for network reconstruction.

Purpose of the Study:

• To develop a novel framework for accurately reconstructing evolutionary game networks.
• To overcome the limitations of parameter selection in existing network reconstruction approaches.
• To provide a robust method for analyzing complex system interactions.

Main Methods:

• Framed network reconstruction as a multiobjective optimization problem (MOP).
• Developed a framework using a multiobjective evolutionary algorithm (MOEA) called MOEANet.
• Incorporated a lasso-based initialization operator and knee-region-based solution selection.

Main Results:

• The MOEANet framework effectively reconstructs various synthetic and real-world EG networks.
• The proposed method successfully avoids the critical parameter selection problem.
• High accuracy in reconstructing evolutionary game network structures was achieved.

Conclusions:

• MOEANet offers a superior approach to evolutionary game network reconstruction.
• The method enhances understanding and control of collective dynamics in complex systems.
• This framework provides a robust solution for analyzing network structures from profit data.