Evolutionary Optimization in Networked Systems
2008년 12월 18일 목요일 PM 4:00
302 Building, Room 105
The recent surge in complex network research has set the stage for a new era in the study of optimisation in complex networks. Optimization of network performance is a common property of both naturally evolved and man-made systems. There is increasing interest on dynamical processes in complex networks and on how the interplay between these processes and network structure influences the performance of the networked system. In this context, there are entire classes of problems, ranging from epidemic spreading to the control of cascading failures, which are naturally defined as optimization problems. Traditionally, optimization has a strict mathematical definition, which refers to obtaining the solution of a well-defined problem. Here we have to adopt a looser definition of the word by extending it to include a tendency of the system to improve its behavior as a result of a selection pressure naturally or artificially imposed. In this sense evolutionary computation provides a unified approach for optimization in complex networks. With evolutionary modeling, complex networks evolve their structure to optimize the multiple fitness functions with network performances and the design cost. We discuss evolutionary optimization may provide a powerful tool for designing the most desirable network topology by trading off multiple performance measures and topological features in a single coherent formalism. This talk focuses on the following topics: # Evolutionary optimization of networks # Application to communication traffic networks # Application to diffusion networks # Application to consensus problems on networks
Prof. Akira Namatame studied Applied Physics at the Japan Defense Academy (NDA) from 1969, graduating with a BS in 1973. He entered the Operations Research program of Stanford University in 1976, graduating MS in 1977. His focus changed slightly for his Ph. D. studies, joining the Department of Engineering - Economic Systems of Stanford University, and graduating in 1979, with a thesis entitled 'Analyses of dynamic competitive systems. Prof. Namatame joined the Department of Computer Science of NDA in 1988. He is now Dean of the Graduate School and a professor of the Department of Computer Science. He has also occupied positions as a visiting Professor at George Mason University (USA) and Chuo University (Japan). Prof. Namatame is well-known as an international research leader in the application of agent and evolutionary modelling technologies to problems in economic and social research, and in the past ten years he has given over 20 invited talks in these areas. His research interests include multi-agent systems, complex networks, evolutionary computation, and game theory. Prof. Namatame is the editor-in-chief of Springer's Journal of Economic Interaction and Coordination. He has published more than 200 refereed scientific papers, together with eight books on multi-agent systems, collective systems and game theory.