2
$\begingroup$

I read some papers on Quantum GANs, for instance this one and this one. I also noticed all the main quantum computing frameworks have a tutorial on quantum GANs, e.g. qiskit.

However I don't really get what are the practical advantages of the quantum version. Is it a matter of speedup of the convergence of the model ? Is it related to the expressiveness of the model ? In general, what can I expect this quantum version to be better at and why should I choose it ?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

Loading Data into quantum circuit is generally a hard problem which by using QGAN we can reduce it's time complexity Quantum Generative Adversarial Networks for learning and loading random distributions. Generally the time complexity of QGAN is $O(ploylog N)$ which exponantialy faster than classical counterpart $O(N)$.

Also there are some use cases in probing the chemical space using thier large hilbert space Quantum Generative Models for Small Molecule Drug Discovery

Improve this answer
$\endgroup$
4
  • $\begingroup$ Can you point me out to an application closer to the field of data science where using the qGANs is indeed better ? I'm a computer scientist exploring this field for the kind of topics I usually work with, then unfortunately a usecase related to chemistry is not helpful for me $\endgroup$
    – mpro
    Commented Jul 6, 2022 at 7:50
  • $\begingroup$ Also, can you please clarify the complexity of what is O(N) vs O(polylog N)? Is it the complexity to perform a training step or maybe the complexity of generating an image ? $\endgroup$
    – mpro
    Commented Jul 7, 2022 at 7:09
  • $\begingroup$ Basically the QGAN introduced in that paper can be tested in other areas, it's a hybrid quantum generator and classical discriminator so there is a flexibility in building the model. The complicity in for 'low rank matrix computation' or training and predicting. $\endgroup$
    – Farhad
    Commented Jul 11, 2022 at 22:51
  • $\begingroup$ Hello, thanks for you answer. I fear I did not understand the last part related to (I guess) the computational complexity $\endgroup$
    – mpro
    Commented Jul 12, 2022 at 8:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.