首页> 外文OA文献 >Simulations of oligopolistic markets with artificial agents: Decision procedures as emergent properties of adaptive learning
【2h】

Simulations of oligopolistic markets with artificial agents: Decision procedures as emergent properties of adaptive learning

机译:用人工代理模拟寡头市场:决策程序作为自适应学习的新兴属性

摘要

While economic models of strategic interaction among autonomous decision-makers are usually based upon principles of optimisation, this work focuses on "satisfying" decision procedures. A flexible simulator of oligopolistic economic environment, where autonomous decision-makers evolve their decision procedures using a learning and adaptation process, has been built. Each artificial agent is implemented using a feedforward neural network. The unsupervised learning of the agent is obtained using genetic algorithms which evolve the structure and weights of the neural network during simulations. The obtained results show that as the complexity of the environment overwhelms the cognitive abilities of the agents, decision procedures emerge that are at the same time simple robust and "satisfying".
机译:尽管自主决策者之间的战略互动的经济模型通常基于优化原则,但这项工作的重点是“令人满意的”决策程序。建立了一个灵活的寡头经济环境模拟器,自治决策者在其中使用学习和适应过程来发展其决策程序。每个人工代理都使用前馈神经网络来实现。代理的无监督学习是使用遗传算法获得的,该算法在仿真过程中会演化神经网络的结构和权重。获得的结果表明,由于环境的复杂性压倒了代理商的认知能力,因此出现了既简单又健壮和“令人满意”的决策程序。

著录项

  • 作者

    Baldassarre Gianluca;

  • 作者单位
  • 年度 1996
  • 总页数
  • 原文格式 PDF
  • 正文语种 en
  • 中图分类

相似文献

  • 外文文献
  • 中文文献
  • 专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号