I want to know about MNIST fpga gui-example.
I care about the expression ‘trained’ in the MNIST example.
I have verified that the weight of each neuron in the argument data file sent to the board is always constant. And in the MNIST example using the fpga board, I also noticed that there is no code to feed back the error like the other examples.
I’m wondering if I’m learning MNIST images on fpga board.
If the board is learning how to learn.
If the board is not learning, is it a model that has already been learned, or is it simply an image that represents a spike?
And I wonder what data is transmitted between PC and board through UDP