Interpretable Optimization Training Strategy-Based DCNN and Its Application on CT Image Recognition

Algorithm 1

The samples are randomly discrete and then input into the model for training.

	Calculate and obtain the number of input training samples:
m = size(x, 3);
Obtain the sample number after random dispersion:
kk = randperm(m);
Get batch sample size:
numbatches = m/opts.batchsize;
for l = 1: numbatches
Take out the batchsize samples and corresponding labels after disordering
batch_x = x(:,:, kk((l - 1) opts.batchsize + 1: l opts.batchsize));
batch_y = y(:, kk((l - 1) opts.batchsize + 1: l opts.batchsize));
Computes the network output under the current network weights and network inputs:
net = cnnff(net, batch_x);
After the above network output is obtained, bp algorithm is used to get the error pair network weight through the corresponding sample label:
net = cnnbp(net, batch_y);
After obtaining the derivative of the error to the weight, the weight is updated by the weight update method:
net = cnnapplygrads(net, opts);
if isempty(net.rL)
net.rL(1) = net.L;
end
net.rL(end + 1) = 0.99 net.rL(end) + 0.01 net.L;
end