An Indexed Bibliography of Genetic Algorithms and Neural Networks

An Indexed Bibliography of Genetic Algorithms in Operations Research

To cope with increasing customer demand and market changes, airline companies need to organize and manage their route networks in a more cost-efficient way. In addition, the robustness of flight operations against unpredictable accidents such as terrorist attacks and natural disasters has become more important to airlines. In this study, the concepts and techniques from complex network theory are used to model airline route networks, and then an effective and efficient genetic algorithm is developed to optimize airline route networks in terms of network properties that may have crucial roles to play in improving the cost-effectiveness and reliability of airspace systems. The simulation results illustrate that the work reported in this study has a good potential to improve the topology of airline route networks in terms of given network properties such as operating costs and network robustness.

Application of Complex Network Theory and Genetic Algorithm in Airline Route Networks

In this paper, a neural network is used to implement an optimized objective function for a genetic algorithm (GA) for application on array antenna design optimization. Traditional GAs are inefficient because a large amount of data that describes the problem space is discarded after each generation. Using the neural network enhanced genetic algorithm (NNEGA), this redundant information is fed back into the GA's objective function via the neural network. The neural network learns the optimal weights of the objective function by identifying trends and optimizing weights depending on the knowledge that it accumulates in-situ. The NNEGA is successfully applied to challenging array antenna design problems. This use of neural network to optimize a multi-objective function for the GA is a new idea that is different from other hybridization of GA and NN.

/pdf/a-novel-evolutionary-learning-technique-for-multi-objective-3921mxabvv.pdf

A novel evolutionary learning technique for multi-objective array antenna optimization

Recently complex network theory has been broadly applied in various domains. How to effectively and efficiently optimize the topology of complex networks remains largely an unsolved fundamental question. When applied to the network topology optimization, Genetic Algorithms (GAs) are often confronted with permutation representation, memory-inefficiency and stochastic modeling problems, as well as difficulties in the design of problem-specific evolutionary operators. This paper, inspired by the natural ripple spreading phenomenon, reports a deterministic model of random complex networks. Unlike existing stochastic models, the topology of a random network can be thoroughly determined by some ripple-spreading related parameters in the new model. Therefore, the network topology can be improved by optimize these ripple-spreading related parameters. As a result, no purpose-designed GA is required, but a very basic binary GA, compatible to all classic evolutionary operators, can be applied in a straightforward way. Preliminary simulation results demonstrate the potential of the proposed ripple-spreading model and GA for the topology optimization of random complex networks.

Hai-jun Zhang

Papers

Neural Network and Genetic Algorithms for Topology Optimization of the CCS7 Network