I have a quick question about "Custom Regression Output Layer."_ https://www.mathworks.com/help/nnet/ug/define-custom-regression-output-layer.html
My goal is similar to this: https://www.mathworks.com/help/nnet/examples/train-a-convolutional-neural-network-for-regression.html
My understanding is that, in the first step, a batch of images goes through the network and compute the loss by learning a function of the forwardloss. However, I realized that MATLAB runs a function of the backwardloss first.
Could you explain why the backwardloss runs first?