A constrained evolutionary computation method for detecting controlling regions of cortical networks.

Area of Science:

• Neuroscience
• Network Science
• Control Theory

Background:

• Cortical networks are crucial for brain function, and identifying controlling regions is key to understanding and manipulating network dynamics.
• Optimal selection of these controlling regions remains a challenge, with existing methods lacking a unified framework for multiple controllability measures.

Purpose of the Study:

• To develop a unified framework for detecting controlling regions in complex networks by integrating two controllability measures: eigenratio (R) and maximum imaginary part of the extended connection matrix (σ).
• To investigate the detection of controlling regions in weighted, directed complex networks, specifically cortical networks, using a novel computational approach.

Main Methods:

• Formulated the detection of controlling regions as a constrained optimization problem (COP), minimizing R while constraining σ.
• Developed and applied an improved dynamic hybrid framework (IDyHF) for solving the COP and identifying controlling regions.
• Validated the IDyHF against evolutionary computation, control theory, and graph theory methods.

Main Results:

• The IDyHF successfully detected controlling regions in a cat cortical network, demonstrating superior performance compared to existing methods.
• Identified that controlling regions are typically characterized by a high in-degree and low out-degree.
• Revealed the dependence of controlling region selection on the number of driver nodes (l) and the constraint level (r), with σ becoming more critical as l increases.

Conclusions:

• The proposed IDyHF offers an effective and unified approach for identifying critical controlling regions in complex networks.
• The findings provide insights into network controllability, applicable to diverse systems like transportation, genetic, and social networks.
• This research advances the understanding and potential manipulation of complex network dynamics for improved coordination and information consensus.