A plug-and-play mechanism that mixes genetic-algorithm evolution into RL training for neural routing solvers gives small benchmark gains, but its stability theorem is not valid as proven.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Synergizing Reinforcement Learning and Genetic Algorithms for Neural Combinatorial Optimization
A plug-and-play mechanism that mixes genetic-algorithm evolution into RL training for neural routing solvers gives small benchmark gains, but its stability theorem is not valid as proven.