针对复杂网络的社区结构挖掘、结构平衡的计算与转换以及结构鲁棒性的行为优化展开研究。建立适用于大规模复杂网络社区结构检测的多层学习模型，提出局部搜索能力强、具备参数自适应学习能力的启发式多层学习Memetic算法，有效挖掘复杂网络拓扑结构知识。建立复杂网络结构平衡性分析模型，挖掘网络社区内部的不平衡因素，实现复杂网络从非平衡结构向平衡结构的转换。构建多层定向攻击模型，建立基于结构鲁棒性增强的启发式网络拓扑结构微调方法，提高复杂网络抵抗外界攻击的能力。此外，分析各模型对算法性能的影响，验证模型和算法的有效性与先进性。

This proposal mainly focuses on the community detection, the structural balance computation and transformation and the structural robustness optimization in complex networks. To detect communities structures from large-scale complex networks, a multi-level learning model will be designed. By establishing heuristic multi-level learning memetic algorithms with enhanced local search and parameter-adaptive learning ability, the project will well discover knowledge of complex networks topologies. By establishing the analysis model of structural balance, unbalanced networks will be converted into balanced ones by transforming the identified unbalanced relationships. With the proposed multi-level target attack models, this project will design structural robustness oriented heuristic methods for changing network topologies which could enhance the abilities of complex networks to resist unpredictable attacks. Besides, the effectiveness and the advantages of the models and algorithms will be validated.