| Summary: | 碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 92 === The social environment changes very fast at every moment. Research and development will be intensified to bring about the enterprise to confront the greater accrual of information. Therefore, Enterprise has to adjust artifice of administration for more new and change. Enterprise promotes overall managerial efficiency from inside to outside and technique service quality in order to shoot competition among marketplace for leading role.
An inner management is lifeblood of enterprise. Under information technology to incite, enterprise is up against no longer the slow step of human management by conventional condition, but is the fastest real time response of information requirement. The control of save quantity is the most important part of enterprise to manage in informational chapter. The purpose of research is how to use the optimum way to control the kind and quantity of material. Expecting to keep the least cost of saving quantity and to exploit the best ability of supply.
This thesis brings up the implementation about classification of material by the plastic perceptron neural network (PPNN). The applied structure of PPNN in thesis is to improve the disadvantage of the lower learning rate, not easy to converge and sample data change because of adding or bating having to retrain. The base institute basic principle of plastic perceptron neural network is to divide the conventional neural network to many subnetwork, and one subnetwork represents one class. All subnetwork are independent and parallel processing. We don’t need to retrain unadorned network on account of adding or bating any one of the subnetwork is permitted. So the PPNN has plastic characteristics, especially applies to use in classification and the discriminating rate of classification is very high.
|
|---|
