Graph-Neural-Network-Based Combinatorial Optimization under Natural Language Soft Constraints
The development of 6G network management technology relies on the solution of large-scale combinatorial optimization (CO) problems on graphs. Many of these problems are NP-hard and must be solved approximately in practice. Despite decades of research and development, classical heuristics and approximation algorithms still struggle with accuracy and the computational demands of scaling to large pro
