The paper intends to introduce you the foundation of Artificial Neural Network(ANN) theory by constructing a neural network for a credit scoring system step by step. And there are three chapters in the paper.
Chapter 1 gives you at first an overview of ANN, the characteristics for ANN. Then MP model is introduced, which is is a multi- inputs and outputs processing cell for nonlinear information and is the foundation for ANN. Afterwards the classical Multilayer Feedforward Neural Network (MFNN) or so called Multilayer Perceptron (MLP) is introduced, which is the unique type of ANN used in the paper. The Back Propagation (BP) training algorithm is a kind of Nonsupervised Learning (NSL), which is very suitable to MLP and is roughly introduced.
Chapter 2 introduces you all the process to construct an MLP for a credit scoring system. At first a description of credit scoring system is given, then construct a primary MLP. Some theories and practices for analysing the data is introduced, which is important for simplifying the network, improving the performance and enhancing the efficiency. Afterwards the structure of MLP is discussed and the MLP for a credit scoring system then can be got.
Summary lists the principles and sums up the experience to build a MLP according to specific requirements.
|
Football
