论文信息 - Agent-based optimization of neural classifiers

Agent-based optimization of neural classifiers

In the paper an agent-based system of evolving neural networks dedicated to solving classification problem is presented. Next, aspects of the system concerning management of collective intelligence and evolution of parameters of neural network are discussed. Evolutionary multi-agent system (EMAS) is described with enhanced immune-inspired selection mechanism. Finally selected results of the experiments are presented.

A. Byrski | J. Dobrowolski | K. Toboła

[1] Athanasios Kehagias,et al. Predictive Modular Neural Networks , 1998 .

[2] Simon Haykin,et al. Neural Networks: A Comprehensive Foundation , 1998 .

[3] C. Kaynak,et al. Techniques for Combining Multiple Learners , 1998 .

[4] Marek Kisiel-Dorohinicki. Agent-Oriented Model of Simulated Evolution , 2002, SOFSEM.

[5] Vassilios Petridis,et al. Predictive Modular Neural Networks: Applications to Time Series , 1998 .

[6] Xin Yao,et al. Evolving artificial neural networks , 1999, Proc. IEEE.

[7] Dorothea Heiss-Czedik,et al. An Introduction to Genetic Algorithms. , 1997, Artificial Life.

[8] Marek Kisiel-Dorohinicki,et al. Collective Intelligence from a Population of Evolving Neural Networks , 2003, IIS.

[9] Marek Kisiel-Dorohinicki,et al. Agent-Based Evolutionary and Immunological Optimization , 2007, International Conference on Computational Science.