First: link and also previous difficulty in transpiling circuit from one library to another.

What it does

Provides as an input the output you want (the number of counts for each combination, say "0000", "0100", etc), then output the rotational values for a predefined gates combination consisting of Rx, Ry, and Rz.

How we built it

Framing the problem as either a supervised learning model or reinforcement learning model to learn the hyperparameters (rotational value) of the gates.

Challenges we ran into

Trying to build a GAN using a qiskit circuit as a discriminator and realizing it is not backpropagate-able. Difficulty in getting the gradient to descent (persist and unsolved) in our model. Difficulty in framing the problem into something the machine easily learns.

Accomplishments that we're proud of

It works!

What we learned

More deeper knowledge in experimenting how to integrate quantum circuits into Machine Learning.

What's next for Tuning of Quantum Circuit Hyperparameters

Integrating with other methods of tuning such as placement of quantum gates tuner, and compiled into a single program will be able to launch for users to use it either in cloud or as an application to get their desired output. Furthermore, we could create gates with this desired output and they could be combined together with other gates made the same way into a complete circuit.

Built With

  • google-colab
  • python
  • qiskit
  • strangeworks
  • tensorflow
Share this project: