-
Notifications
You must be signed in to change notification settings - Fork 7
/
launchNetwork.m
34 lines (28 loc) · 1.06 KB
/
launchNetwork.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
function [net] = launchNetwork(data_x, target_y)
% This script makes returns a trained CNN given training data
%
% Input paramters (arguments) are:
% data_x: 4-D double containing the input data in the following form
% [128 1 1 numberOfReadings]
% target_y: 1-D Categorial containing the signal labels
% [numberOfReadings 1]
% Output values returned are:
% net: trained neural network
% Author: 06/15/17 - by Arshan Hashemi
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
layers = [imageInputLayer([128 1], 'Normalization', 'none')
convolution2dLayer([15 1],32)
reluLayer
maxPooling2dLayer([4,1]);
convolution2dLayer([15 1],16)
reluLayer
maxPooling2dLayer([2,1]);
dropoutLayer();
fullyConnectedLayer(12);
fullyConnectedLayer(6);
softmaxLayer
classificationLayer()];
options = trainingOptions('sgdm','MaxEpochs',50, ...
'InitialLearnRate',0.001);
rng('default')
net = trainNetwork(data_x, target_y, layers, options);