Intelligent Systems
Here's a part of the work done by Anish and me on the ANN and some other relevant scripts, results.
Artificial Neural Network(1).docx has brief explanations of the code and results.
Artificial_neural_network_1.docx
Layout for methodology-
Codes
mintomat.m --> mathematical analysis of peak pressure to determine suitable insole
inputmaker.m --> prepares input matrices for other scripts and returns these when called
ANNextraneuron.m --> 3 layered ANN, trains to converge to a value of 0.7 for the given training set
trainANDtest.m --> trains the ANN using inputmaker.m and uses 3 duplicate copies of inputmaker.m to test the neural network for three sets of data simultaneously.
density.m --> mathematical analysis of peak pressure to determine suitable DENSITY for a given insole material. This code assumes that density is a subset of material and therefore considers pressure values for only the material recommended by the previous code.
Note- The comments in these scripts may need to be updated and might even seem irrelevant at some places. Please ignore these for now.
Graphs for outputs and mean square error
The "graphs_ANN_material" document has some results which have been plotted as graphs (output value vs iterations, MSE vs iterations, RMSE vs iterations). Please see Artificial Neural Network(1).docx to understand significance of these results.