08-01-2014, 03:08 PM
Graph Based Genetic Algorithm as Evolutionary Computing
Abstract:
Graphed Evolutionary Computing is a generic and flexible way to model connected evolutionary algorithms with the use of directed graphs. The flexible nature of GEC makes it possible to model different types of parallel evolutionary computing, including mixtures of different evolutionary computing and its settings of parameters. A tree shaped graphed genetic algorithm (TGGA) is designed as an example of applied GEC. Its bottom-up information flow exists out of a set of solutions (local optima), which TGGA successfully combines and mixes. This makes TGGA a very effective parallel GA. The LAM/MPI implementation of TGGA can be executed on a heterogeneous local area multi-computer [1] in an asynchronous way and has low communication bandwidth requirements
Abstract:
Graphed Evolutionary Computing is a generic and flexible way to model connected evolutionary algorithms with the use of directed graphs. The flexible nature of GEC makes it possible to model different types of parallel evolutionary computing, including mixtures of different evolutionary computing and its settings of parameters. A tree shaped graphed genetic algorithm (TGGA) is designed as an example of applied GEC. Its bottom-up information flow exists out of a set of solutions (local optima), which TGGA successfully combines and mixes. This makes TGGA a very effective parallel GA. The LAM/MPI implementation of TGGA can be executed on a heterogeneous local area multi-computer [1] in an asynchronous way and has low communication bandwidth requirements