Gradient descent backpropagation
net.trainFcn = 'traingd'
sets the networktrainFcn
property.
[
火车的network withtrainedNet
,tr
] = train(net
,...)traingd
.
traingd
is a network training function that updates weight and bias values according to gradient descent.
Training occurs according totraingd
training parameters, shown here with their default values:
net.trainParam.epochs
— Maximum number of epochs to train. The default value is 1000.
net.trainParam.goal
— Performance goal. The default value is 0.
net.trainParam.lr
— Learning rate. The default value is 0.01.
net.trainParam.max_fail
— Maximum validation failures. The default value is6
.
net.trainParam.min_grad
— Minimum performance gradient. The default value is1e-5
.
net.trainParam.show
— Epochs between displays (NaN
for no displays). The default value is 25.
net.trainParam.showCommandLine
— Generate command-line output. The default value isfalse
.
net.trainParam.showWindow
— Show training GUI. The default value istrue
.
net.trainParam.time
— Maximum time to train in seconds. The default value isinf
.
traingd
can train any network as long as its weight, net input, and transfer functions have derivative functions.
Backpropagation is used to calculate derivatives of performanceperf
with respect to the weight and bias variablesX
. Each variable is adjusted according to gradient descent:
dX = lr * dperf/dX
Training stops when any of these conditions occurs:
The maximum number ofepochs
(repetitions) is reached.
The maximum amount oftime
is exceeded.
Performance is minimized to thegoal
.
The performance gradient falls belowmin_grad
.
Validation performance has increased more thanmax_fail
times since the last time it decreased (when using validation).