b. One of the steps in designing a neural network is training steps through a process...

Transcribed Image Text:

b. One of the steps in designing a neural network is training steps through a process called backpropagation process. To understand the backpropagation process, run the following Matlab program. clc close all warning off net-newff ([-2 2; -2 -2], [2 1], ['logsig' logsig'},'trainlm'); net.b(2)[0] net.b{1} = [0,0] net. IW (1,1} =[0.1 0.4;0.8 0.6] net.LW {2,1} [0.3 0.9] weights = getwb (net) p=[0.35 0.9] a = sim (net.p.) = i. Analyse all parameters of the neural network generated by the program and calculate the output error of the network. 1 ii. To reduce the error a backpropagation using gradient descent method was performed and the new weight for input and output layers are: [0.0992 0.39721 0.7978 0.5928] [0.272 0.873] Test the new weight to the network and comment on error improvement. iii. Illustrate the backpropagation process with new weight that will give better error result b. One of the steps in designing a neural network is training steps through a process called backpropagation process. To understand the backpropagation process, run the following Matlab program. clc close all warning off net-newff ([-2 2; -2 -2], [2 1], ['logsig' logsig'},'trainlm'); net.b(2)[0] net.b{1} = [0,0] net. IW (1,1} =[0.1 0.4;0.8 0.6] net.LW {2,1} [0.3 0.9] weights = getwb (net) p=[0.35 0.9] a = sim (net.p.) = i. Analyse all parameters of the neural network generated by the program and calculate the output error of the network. 1 ii. To reduce the error a backpropagation using gradient descent method was performed and the new weight for input and output layers are: [0.0992 0.39721 0.7978 0.5928] [0.272 0.873] Test the new weight to the network and comment on error improvement. iii. Illustrate the backpropagation process with new weight that will give better error result

Answer rating: 100% (QA)

Youre presented with a MATLAB program designed for a basic neural network and there are three questions regarding it The neural network is defined using the newff function which creates a feedforward ... View the full answer