Co-evolutionary Multi-agent Learning by means of Genetic Programming

通过遗传编程的协同进化多智能体学习

• 批准号: 11480071
• 负责人: IBA Hitoshi
• 金额: $ 4.61万
• 依托单位: UNIVERSITY OF TOKYO
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (B).
• 财政年份: 1999
• 资助国家: 日本
• 起止时间: 1999 至 2000
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-11480071/
• 关键词: Evolutionary System; Multi-agent System; Genetic Algorithms; Genetic Programming; Machine Learning; Co-evolution; Symbiosis; マルチエージェント

It is not easy to describe in advance the program which controls more than one agent effectively. This is due to the difficulty with multi-agent programming for the communication, cooperation, synchronization and conflict resolution. Thus, we propose a multi-agent learning scheme based on the co-evolutionary method for the purpose of the automatic generation of the programs for many agents. Our approach is based upon the evolutionary computation, i.e., genetic programming and genetic algorithms.The established multi-agent learning system has been applied to several practical problems, and the validity has been shown. The application areas include the wider field of examples as described below :(1) Adaptive WWW search by the Web robot.(2) Automatic synthesis of analog circuits, such as an asymmetric bandpass filter.(3) Inference of gene regulatory networks, i.e., Weaver models or differential equation systems, for genome informatics.(4) Interactive music composition.(5) Evolutionary quantum circuit design, such as quantum teleportation and oracle problems.(6) Financial data prediction, such as Nikke225 or foreign exchange rate.

预先描述有效控制多个智能体的程序并不容易。这是由于多智能体编程在通信、合作、同步和冲突解决方面存在困难。因此，我们提出了一种基于共同进化方法的多智能体学习方案，以自动生成多个智能体的程序。我们的方法基于进化计算，即遗传编程和遗传算法。所建立的多智能体学习系统已应用于多个实际问题，并显示了有效性。应用领域包括更广泛的示例领域，如下所述：（1）网络机器人的自适应万维网搜索。（2）模拟电路的自动合成，例如非对称带通滤波器。（3）基因调控网络的推理，即用于基因组信息学的韦弗模型或微分方程系统。（4）交互式音乐创作。（5）进化量子电路设计，例如量子 (6)金融数据预测，如Nikke225或外汇汇率。

项目成果

Hitoshi IBA,Nikolay I.Nikolaev 他: "Inductive Genetic Programming with Immure Network Dynamics"Advances in Genetic Programming. Vol.3. 355-376 (1999)

Hitoshi IBA、Nikolay I. Nikolaev 等人：“具有不固定网络动力学的归纳遗传编程”遗传编程进展，第 3 卷（1999 年）。

Ando, S., Iba, H.: "Analog Circuit Design with a Variable Length Chromosome"J.Japan Artificial Intelligence Society. vol.15, no.5. 844-853 (2000)

Ando, S., Iba, H.：“可变长度染色体的模拟电路设计”J.日本人工智能学会。

Iba, H.: "Evolving Multiple Agents by Genetic Programming"Genetic Programming 3, Spector, L., Langdon, W.B., O'Reilly, U.-A., and Angeline, P.A., (eds.), MIT Press. 447-466 (1999)

Iba, H.：“通过基因编程进化多个智能体”《基因编程 3》，Spector, L.、Langdon, W.B.、OReilly, U.-A. 和 Angeline, P.A.（编辑），麻省理工学院出版社。

伊庭斉志: "遺伝的プログラミング"電子情報通信学会誌. vol.83,no.5. 251-258 (2000)

Hitoshi Iba：《基因编程》电子、信息和通信工程师学会杂志，第 83 卷，第 5 期（2000 年）。

Ito, T., Iba, H., Sato, S.: "A Self-Tuning Mechanism for Depth-Dependent Crossover"Genetic Programming 3, Spector, L., Langdon, W.B., O'Reilly, U.-A., and Angeline, P.A., (eds.), MIT Press. 377-399 (1999)

Ito, T.、Iba, H.、Sato, S.：“深度依赖交叉的自调节机制”遗传编程 3、Spector, L.、Langdon, W.B.、OReilly, U.-A.，