What does a value of performance in a neural network indicate?

performance = perform(net,Out,y)

trainTargets = Out .* tr.trainMask{1};

valTargets = Out .* tr.valMask{1};

testTargets = Out .* tr.testMask{1};

trainPerformance = perform(net,trainTargets,y)

valPerformance = perform(net,valTargets,y)

testPerformance = perform(net,testTargets,y)

The above is the code I use to calculate training,validation and testing performace. I get values like :

performance =0.9892

trainPerformance = 0.7206

valPerformance =0.5901

testPerformance = 0.5925

what do these values indicate? What are the range of values I should aim at? I have a 603x11 data set.

Thanks in advance.

Answer 1

[gonzalo Mier]

If you look in the documentation of perform function (https://www.mathworks.com/help/deeplearning/ref/perform.html) it tells you "returns network performance calculated according to the net.performFcn and net.performParam property values."

Also, looking in the documentation of Neural Network Object Properties, you can look those parameters and how to change them (https://www.mathworks.com/help/deeplearning/ug/neural-network-object-properties.html#bss4hk6-52)