After pooling layer base implementation issue being resolved, feedforwarding function and backpropagating function are left to be given shape to. feedForward() will resize input matrix using max pooling or average pooling. backPropagate() would compute current error using next error. The computation function would be different depending on pooling types. A few matrix computation functions needed to implement two functions above will be added, too.
After pooling layer base implementation issue being resolved, feedforwarding function and backpropagating function are left to be given shape to.
feedForward()
will resize input matrix using max pooling or average pooling.backPropagate()
would compute current error using next error. The computation function would be different depending on pooling types. A few matrix computation functions needed to implement two functions above will be added, too.