After some nonlinear activation function.y = ?(?ni=0wixi+b)=?(WTX+b) (1)where

After the data has been processed and transformed ,methodof training has been chosen, it is necessary to then determinethe topology of the neural network. The networktopology describes the arrangement of the neural network 3.Choosing the topology of the neural network is a difficultdecision 4.There are different types of network.We canselect as our demand.Each network provides some advantagesand disadvantages.For example, some networks are good forspeed accuracy, while some of them are good for handlingstatic variables and not continuous ones. Hence, form variousnetwork topologies such as Multilayer Perceptron, recurrentnetwork,and time-lagged recurrent network were considered.Due to the nature of our work, the Multilayer Perceptronwas selected. An MLP is a network of simple neurons calledperceptrons.The perceptron computes a single output from multiple real-valued inputs by forming a linear combinationaccording to its input weights and then possibly putting theoutput through some nonlinear activation function.y = ?(?ni=0wixi+b)=?(WTX+b) (1)where w denotes the vector of weights, x is the vector ofinputs, b is the bias and ? is the activation function.X = f(s) = B?(As + a) + b (2)where s is a vector of inputs and x a vector of outputs. A isthe matrix of weights of the first layer, a is the bias vectorof the first layer. B and b are, respectively, the weight matrixand the bias vector of the second layer. The function ? denotesan element wise non linearity. A typical multilayer perceptron(MLP) network consists of a set of source nodes forming theinput layer, one or more hidden layers of computation nodes,and an output layer of nodes. The input signal propagatesthrough the network layer-by-layer6.It is so tough building aneural network model with number of nodes and hidden layers.Because small number of hidden layer lower the processingcapability.Comparatively the system will slow down if a largenumber of hidden layer.We come to a conclusion from thisparadigm ,choose a network with a hidden layer and eightinput processing elements, one output.Train this network withdifferent learning rate and different hidden layer.For whichlarning rate and hidden layer our prediction came closer wetook this network.