Komparasi Algoritma Multilayer Perception dan Support Vector Machine dalam Pemilihan Beasiswa

Abstract

This section introduces multilayer perceptrons, which are the most commonly used type of neural network. The popular backpropagation training algorithm is studied in detail. The momentum and adaptive step size techniques, which are used for accelerated training, are discussed. Other acceleration techniques are briefly referenced. Several implementation issues are then examined. The issue of generalization is studied next. Several measures to improve network generalization are discussed, including cross validation, choice of network size, network pruning, constructive algorithms and regularization. Recurrent networks are then studied, both in the fixed point mode, with the recurrent backpropagation algorithm, and in the sequential mode, with the unfolding in time algorithm. A reference is also made to time-delay neural networks. The section also includes brief mention of a large number of applications of multilayer perceptrons, with pointers to the bibliography. C1.2.1 Introduction Multilayer perceptrons (MLPs) are the best known and most widely used kind of neural network. They are formed by units of the type shown in figure C1.2.1. Each of these units forms a weighted sum of its inputs, to which a constant term is added. This sum is then passed through a nonlinearity, which is often called its activation function. Most often, units are interconnected in a feedforward manner, that is, with B3.2.4 interconnections that do not form any loops, as shown in figure C1.2.2. For some kinds of applications, recurrent (i.e. nonfeedforward) networks, in which some of the interconnections form loops, are also used.