Forward propagation in Deep Neural Networks

In the "Forward Propagation in a Deep Network" video on Coursera, Andrew NG mentions that there's no way to avoid a for loop to loop through the different layers of the network during forward propagation.

See image showing a deep network with 4 layers, and the requirement of a forloop to compute activations for each layer during forward propagation: https://nimb.ws/CkRVLT

This makes intuitive sense since each layer's activation depends on the previous layer's output.

Warning: start of speculation

My rudimentary understanding of quantum computing is that it somehow "magically" can bypass computing intermediate states -> this is why supposedly quantum computers can break cryptography... or something like that.

I'm wondering if a quantum computer could perform vectorized forward propagation on an L layer deep neural network.

  • 1
    $\begingroup$ just commenting to point out the existence of quantumcomputing.SE, in case you were not aware of the site $\endgroup$
    – glS
    Jul 13, 2020 at 15:49


You must log in to answer this question.